Impaired development of memory B cells and antibody responses in humans and mice deficient in PD-1 signaling

Summary

T follicular helper (Tfh) cells abundantly express the immunoreceptor PD-1, and the impact of PD-1 deficiency on antibody (Ab)-mediated immunity in mice is associated with compromised Tfh cell functions. Here, we examined the role of the PD-1:PD-L1 axis on Ab-mediated immunity. Individuals with inherited PD-1 or PD-L1 deficiency had fewer memory B cells and impaired Ab responses, similar to Pdcd1^−/− and Cd274^−/−Pdcd1lg2^−/− mice. PD-1- or PD-L1-deficient B cells had reduced expression of the transcriptional regulator c-Myc and c-Myc-target genes. A fraction of human B cells expressed PD-1 and PD-L1, and PD-1 deficiency or neutralization of PD-1 or PD-L1 impeded c-Myc expression and Ab production in human B cells in vitro. Furthermore, B-cell-specific deletion of Pdcd1 prevented the physiological accumulation of memory B cells in mice. Thus, PD-1 shapes optimal B cell memory and Ab-mediated immunity through B cell-intrinsic and -extrinsic mechanisms, indicating that B-cell dysregulation contributes to infectious and autoimmune complications following anti-PD-1:PD-L1 immunotherapy.

eTOC blurb

Ogishi, Kitaoka et al. examine humans deficient in PD-1 and PD-L1 as well as associated mouse models and find that that PD-1 and PD-L1 promote B cell memory and antibody responses through both T cell-dependent and B cell-intrinsic mechanisms. Their findings provide insight into the mechanisms underlying the infectious and autoimmune complications associated with anti-PD-1:PD-L1 immunotherapy.

Graphical Abstract

Introduction

Human programmed cell death protein 1 (PD-1, encoded by PDCD1)) is an inhibitory receptor weakly expressed on resting T, B, NK cells, and some myeloid cell subsets, and its expression is increased on T cells following activation ¹. Interactions between PD-1 and its ligands PD-L1 (CD274) and PD-L2 restrain prolonged activation of PD-1-expressing cells. Neutralizing antibodies (nAbs) against PD-1 and PD-L1, which are widely used in cancer immunotherapy, unleash T cell-mediated immunity to self- and tumor-derived antigens (Ags) ². However, the physiological roles of PD-1, PD-L1, and PD-L2 in B cell-mediated immune responses remain incompletely understood. In vitro, PD-1 engagement inhibits activation, proliferation, and Ab production, while PD-1 blockade enhances proliferation of human or murine B cells ^3–6. Nevertheless, Pdcd1^−/−, Cd274^−/−, and Cd274^−/−Pdcd1lg2^−/− mice have fewer total and Ag-specific long-lived plasma cells (PCs) in the spleen and bone marrow (BM) ^7,8. Pdcd1^−/− mice also produce IgA with reduced affinity for commensal bacteria and abnormally high rates of plasmablast apoptosis in mucosal tissues ⁹. PD-1 signaling has a B cell-intrinsic role in PC development ⁷ and a B cell-extrinsic role in regulating IL-21 production - a cytokine crucial for B-cell differentiation - by T follicular helper (Tfh) cells ^7,9,10. These observations suggest PD-1 signaling is essential for shaping optimal Ab responses in vivo by orchestrating the differentiation and function of B and Tfh cells.

Indeed, evidence suggests that an adverse consequence of PD-1:PD-L1-mediated cancer immunotherapy in humans is an increased frequency of bacterial infections, irrespective of treatment with other immunosuppressive reagents ^11–14. Microbes reported to cause disease in the context of PD-1 blockade include S. pneumoniae and H. influenzae, which are typically controlled by Abs, as revealed by studies of inborn errors of immunity affecting human B cells ^15,16. Moreover, anti-PD-1 immunotherapy reduced the titer and affinity of anti- influenza Ab before and after vaccination ¹⁷. However, the underlying cancer and comorbidities of these patients render it challenging to connect the temporal deficit of PD-1 signaling causally and mechanistically with impaired Ab-mediated immunity in humans.

We recently described two Turkish siblings with inherited complete PD-1 deficiency ¹⁸, and two Moroccan siblings with inherited complete PD-L1 deficiency ¹⁹. All affected individuals had a history of early-onset endocrine autoimmunity, including type 1 diabetes. However, the PD-1-deficient siblings succumbed to autoimmune pneumonitis at the ages of 3 and 11 years, while the two PD-L1-deficient siblings are alive and well at ages 10 and 11 years. Although none of these individuals had any history of unusually severe bacterial infections, typically present in patients with Ab deficiency, we decided to exploit this unique opportunity to unravel PD-1:PD-L1-dependent mechanisms regulating human Ab responses by investigating the Ab repertoire and development and function of B cells and Tfh cells in PD-1-deficient and PD-L1-deficient patients.

Results

Human PD-1:PD-L1 axis promotes B-cell memory and Ab responses

We investigated whether PD-1 deficiency compromised humoral immune responses ¹⁸. While plasma levels of IgG1, IgG2, IgG3, IgG4, IgA, IgM, IgD, and IgE samples (obtained at the ages of 10 and 11 years) were normal, overall reactivity of serum IgG from the PD-1-deficient patient to various common microbes (e.g., S. pneumoniae, rhinovirus B) was reduced compared to IgG from healthy relatives and age-matched healthy donors, being similar to patients with combined immunodeficiency due to CARMIL2 deficiency ^20,21 (Fig. 1A, B, S1A). Within the peripheral B-cell compartment of the patients, the proportion of B cells was higher in the PD-1-deficient patient (obtained at 3 different times) than healthy relatives, whereas PD-L1-deficient patients had B-cell proportions similar to healthy donors (Fig. 1C, G). PD-1- and PD-L1-deficient patients had lower proportions of memory B cells and plasmablasts than healthy donors (Fig. 1D–F, H). Moreover, frequencies of IgG⁺ memory B cells were lower in the PD-1-deficient patient and one PD-L1-deficient patient than healthy donors, whereas IgA⁺ B cells were unaffected (Fig. 1I). Finally, we observed high proportion of CD21^loCD23^loT-bet⁺ B cells in the PD-1-deficient, but not PD-L1-deficient, patients, similar to patients with heterozygous STAT3 gain-of-function (GOF) variants (Fig. 1J–L). As described previously ²², T-bet deficiency abolished the development of this subset (Fig. 1J–L). These observations suggest the human PD-1:PD-L1 axis promotes the memory B cell development and effective Ab responses to common environmental microbes in vivo.

Figure 1. Impaired B-cell memory and antibody responses in children with inherited PD-1 or PD-L1 deficiency.

(A-B) VirScan analysis for serum IgG samples from the PD-1-deficient patient (aged 10 and 11 years; two sampling points), his healthy brother (aged 6 and 7 years) and parents, five healthy age-matched controls and 13 patients with biallelic CARMIL2 mutations.

(A) Principal component analysis (PCA). Left: sample distribution. Right: variable loadings.

(B) Scores for representative microbes.

(C-D) Mass cytometry showing (C) B cells among total leukocytes and (D) memory B cells.

(E-L) PBMCs from the PD-1-deficient patient (aged 10 and 11 years; two sampling points), two PD-L1-deficient patients (aged 11 and 10 years), their healthy relatives, IEI controls, and healthy adults and age-matched controls were analyzed for (E) memory B cells and (F) plasmablasts. Graphs depict proportions of (G) B cells among total leukocytes, (H) B-cell subsets, (I) IgA⁺ and IgG⁺ memory B cells.

(J) UMAP representation of B-cell flow cytometry data.

(K) CD21 and T-bet expression in B cells.

(L) CD21^loCD23^loT-bet⁺ B cells.

Graphs represent mean ± SEM. * P < 0.05; *** P < 0.001; **** P < 0.0001.

Reduced BCR repertoire diversity and somatic hypermutation in PD-1 and PD-L1 deficiencies

We assessed the impact of PD-1 and PD-L1 deficiencies on the BCR repertoire by performing bulk RNA sequencing (RNASeq) on (1) naïve and memory B cells from the PD-1-deficient patient (aged 11 years) and his healthy brother (aged 6 years) and (2) whole-blood leukocytes from the PD-L1-deficient patients (aged 11 and 10 years). Notably, the PD-1- and PD-L1-deficient patients had lower CDR3 clonotype diversity for IGHV, IGKV, and IGLV repertoires than healthy adults and age-matched controls (Fig. S2A–C). Somatic hypermutation (SHM) - defined by nucleotide substitutions, insertions, and deletions, most evident in the IGHV CDR1, CDR2, and FR3 – was also lower in B cells of the PD-1-and PD-L1-deficient patients compared to healthy donors (Fig. S2D). This finding is similar to memory B cells from STAT3 GOF and NFKB1 haploinsufficient (HI) patients, who also develop humoral immune dysregulation (Fig S2D). Finally, one PD-L1-deficient patient had low levels of IGHA and IGHG expression, whereas levels in the PD-1-deficient patient and the other PD-L1-deficient patient were at the lower end of the range for healthy donors (Fig. S2E). Inherited PD-1 and PD-L1 deficiencies, therefore, compromise the formation of a diverse, class switched, and somatically mutated BCR repertoire.

Impaired c-Myc target gene expression in PD-1- and PD-L1-deficient B cells

We explored the molecular basis of impaired maturation of PD-1- and PD-L1-deficient memory B cells by analyzing scRNASeq datasets derived from PBMCs from the patients, their healthy relatives, and healthy adult and age-matched controls ^18,19. Single-cell analysis (excluding plasmablasts) from all donors identified clusters representing naïve (expressing IGHD) and memory (IGHA and IGHG) cells (Fig. 2A–C). Clustering analyses confirmed the paucity of transcriptionally defined memory B cells in PD-1- and PD-L1-deficient patients (Fig. 2D), consistent with flow-based immunophenotyping (Fig. 1H). PD-1- and PD-L1-deficient total, naïve, and memory B cells exhibited downregulation of c-Myc target genes (EIF4A1, CDK2, HDAC2) and MYC itself relative to his healthy brother (age 6 years) and healthy age-matched controls (Fig. 2E–H, Fig. S1E). PD-1-deficient B cells also displayed downregulation of NF-κB-driven genes (Fig. 2E). Transcription factor (TF) activity inference analysis also showed a decrease in the predicted activity of TFs that regulate MYC mRNA synthesis (NF-κB1, RelA, c-Rel, STAT3) in PD-1-deficient naïve and memory B cells (Fig. 2I, J). Finally, analysis of a published RNASeq dataset for PD-1-treated cancer patients ¹⁷ revealed downregulation of c-Myc target genes in resting memory B cells (Fig. 2K, L) and Ab-secreting cells following influenza vaccination (Fig. 2M). These findings, therefore, indicate that inherited PD-1 and PD-L1 deficiencies impair c-Myc-driven transcriptional programs in human memory B cells.

Figure 2. Human PD-1:PD-L1 axis enhances c-Myc-driven transcriptional signatures in B cells.

(A-E) scRNASeq analysis for naïve and memory B cells in PBMCs from the PD-1-deficient patient, a healthy brother (aged 10 and 6 years), PD-L1-deficient patients (aged 11 and 10 years), and healthy adult and age-matched controls. (A) UMAP representation of B cells from all donors. Color codes indicate cluster assignments. (B) Top 10 marker genes for each cluster. (C) Genotypes. (D) Memory B cells defined by clustering (mean ± SEM). (E) Geneset overrepresentation analysis.

(F-J) Transcriptomic profiling of naïve and memory B cells sorted from PD-1-deficient patient (aged 11 years), his healthy brother (aged 6 years), age-matched controls, through bulk RNASeq. (F) Representative FACS plots. (G) PCA of global transcriptomic profiles. (H) MA plot. Significantly downregulated c-Myc-targeted genes (defined by Hallmark c-Myc target v1 geneset) are shown in blue. (I, J) Transcription factor activity inference. (I) PCA. (J) Normalized activity of 21 TFs known to regulate MYC mRNA levels among the top 30 TFs according to the PC1 loading in (I).

(K-M) Geneset enrichment analysis (GSEA) of memory B cells and Ab-secreting cells in cancer patients on PD-1 blockade. (K) Memory B cells at baseline. (L) Leading-edge genes for the c-Myc target geneset in (K). (M) Responses to influenza vaccine in Ab-secreting cells.

Impaired memory B-cell development and primary and recall Ab responses in mice deficient for PD-1 signaling

We then assessed central and peripheral B-cell compartments in unimmunized Pdcd1^−/− C57BL/6 (B6) mice ⁴. Young Pdcd1^−/− mice displayed splenomegaly (Fig. S3A), while both young (3–4 months) and aged (11–12 months) Pdcd1^−/− mice had higher B-cell counts in the BM than wild-type (WT) mice (Fig. 3A). These observations are similar to the PD-1-deficient patient ¹⁸. Young and aged Pdcd1^−/− mice also had higher total B-cell counts in lymph nodes (LN) and spleen, respectively (Fig. 3A). However, we found lower proportions of: i) memory (CD73⁺CD80⁺) B cells ²³ in the blood, LNs, spleen, and BM; ii) Ki-67⁺ memory B cells in the BM; and iii) germinal center (GC; CD38^loFas⁺) B cells – both dark (CXCR4^hiCD86^loCD83^lo) and light zone (CXCR4^loCD86^hi CD83^hi) ²⁴ – in the spleen of aged Pdcd1^−/− mice than in WT mice (Fig. 3B–C, S3B–I). Finally, we assessed transcriptomic profiles of naïve (CD73⁻) and memory (CD73⁺CD80⁺) B cells sorted from LNs or spleens of aged (11 months old) Pdcd1^−/− B6 mice. On PCA, the genotype-dependent (i.e., WT vs. Pdcd1^−/−) difference was captured by PC2 for memory B cells and, to a lesser extent, naïve B cells (Fig. 3D). GSEA revealed downregulation of c-Myc and E2F target genes, and G2/M checkpoint-related genes in Pdcd1^−/− LN and splenic memory B cells, but not in naïve B cells, whereas Pdcd1^−/− LN naïve B cells displayed decreased expression of NF-κB-driven genes (Fig. 3E, F). These data confirm that PD-1 promotes the generation of memory B cells in mice and humans.

Figure 3. Impaired B-cell memory and antibody responses in mice deficient for PD-1 signaling.

(A-F) Analysis in young (3–4 mo) and aged (11–12 mo) unimmunized Pdcd1^−/− C57BL/6 (B6) mice. (A) Absolute B-cell counts. (B) Proportions of memory B cells among total B cells (n=5 mice per group). (C) Ki-67 Expression by BM memory B cells. N=5 and 7 for young and aged mice for each genotype. (D-F) Bulk RNASeq analysis in aged Pdcd1^−/− B6 mice. N=4 mice for the spleen. For LNs, cells from three mice were pooled due to the small number of cells available. (D) PCA. (E) GSEA. NES, normalized enrichment score. (F) Heatmap of normalized mRNA levels for leading-edge genes for Hallmark c-Myc target v1 geneset in (E).

(G, H) Impaired development of B-cell memory after oral exposure to bacterial antigens. N=7 and 5 for WT and Pdcd1^−/− mice, respectively. % of (G) memory B cells and (H) IgM⁻IgA⁺ B cells.

(I-L) Impaired Ab responses in mice with impaired PD-1 signaling (I) 12–15 weeks after primary immunization, or at the indicated times following the second immunization in (J) Pdcd1^−/− B6 mice, (K) Pdcd1lg2^−/− B6 mice, (L) Cd274^−/−Pdcd1lg2^−/− BALB/c mice. Results from at least two independent experiments are compiled.

Graphs (A-C, G-H, and J-K) depict mean ± SEM. In (I-L), numbers alongside graphs indicate the number of mice tested. In (I and L), dots represent outliers, as defined by 1.5 times the interquartile range (IQR) above the third quartile. **** P < 0.0001; ** P < 0.01; * P < 0.05; ns, not significant.

We extended our investigation of the role of PD-1 signaling in Ab responses following Ag exposure in mice. We first tested B-cell responses to heat-killed microbial lysates added to the drinking water in young WT and Pdcd1^−/− mice. Orally administered microbial Ag would be captured primarily in the spleen. As expected, proportions of splenic memory and IgM⁻IgA⁺ B cells were lower in Pdcd1^−/− mice than in WT mice (Fig. 3G, H). Good-Jacobson et al. also reported impaired formation of memory B cells and long-lived PCs in PD-1 deficient mice after immunization with specific Ag ⁷. We then tested the role of PD-1 signaling for recall Ab responses after booster immunization in Pdcd1^−/− B6 mice ⁴, Pdcd1lg2^−/− B6 mice ²⁵, and Cd274^−/−Pdcd1lg2^−/− BALB/c mice ²⁶ (Fig. S3J). As previously reported ⁷, Pdcd1^−/− mice had fewer splenic and BM NP-specific IgG1⁺ PCs than WT mice before the second immunization (Fig. 3I). Six days after boosting, NP⁺IgG1⁺ PC numbers remained lower in the BM of Pdcd1^−/− mice compared to WT mice (Fig. 3J). While Pdcd1lg2^−/− mice also had fewer NP⁺IgG1⁺ BM PCs than WT mice after the primary immunization (Fig. 3K, day 0), PD-L2 deficiency alone did not impede the recall response (Fig. 3K). Unlike the BM phenotype, Ag-specific PC numbers were similar in spleens of WT, Pdcd1^−/−, and Pdcd1lg2^−/− mice following the second immunization (Fig. 3J, S3K). Importantly, Cd274^−/−Pdcd1lg2^−/− BALB/c mice phenocopied Pdcd1^−/− B6 mice, demonstrated by low counts of NP⁺IgG1⁺ PCs in the BM, but not spleen, 7 days after the second immunization (Fig. 3L, S3L). Thus, PD-1 signaling governs optimal B-cell memory and primary and recall Ag-specific Ab responses.

Impaired IL-21 expression by PD-1-deficient, but not PD-L1-deficient, memory CD4⁺ T cells

Tfh cells provide critical signals, including IL-21, to promote B-cell proliferation, Ig class switching, and differentiation into memory and PCs ¹⁰. We explored whether human PD-1 and PD-L1 deficiencies compromised Tfh cells. Flow cytometric immunophenotyping showed no difference in the numbers of circulating Tfh (cTfh) cells in peripheral blood of the PD-1- (three experiments on cells sampled at different time points) or PD-L1-deficient (N=2) patients, relative to healthy age-matched and adult controls, and the healthy brother of the PD-1-deficient patient (Fig. 4A, S4A–B). The PD-1-deficient patient, but neither PD-L1-deficient patient, had a modest but higher proportion of CXCR3⁺CCR6⁻ Tfh1-type cells, which are less able to help naïve B cells produce Ig than CXCR3⁻CCR6⁻ Tfh2-type or CXCR3⁻CCR6⁺ Tfh17-type cells ^27,28, than his brother and other healthy age-matched and adult controls (Fig. 4B, S4C). Levels of IL21 mRNA were reduced in activated leukocytes from PD-1-deficient, but not PD-L1-deficient, leukocytes compared to age-matched controls and the healthy brother of the PD-1-deficient patient (Fig. 4C, S4D). Beyond total leukocytes, stimulated PD-1-deficient memory CD4⁺ αβ T cells failed to produce detectable levels of IL-21 (Fig. 4D). Impaired expression of IL21 mRNA and IL-21 protein persisted even after two weeks of in vitro expansion of PD-1-deficient memory CD4⁺ αβ T cells (Fig. 4E, S4G). Lastly, in vitro IL-21 production by activated memory CD4⁺ T cells from aged Pdcd1^−/− mice was impaired (Fig. 4F, S4H–J). Thus, PD-1 – but not PD-L1 – deficiency profoundly impairs the ability of memory CD4⁺ T cells to produce IL-21.

Figure 4. Analysis of Tfh cell-dependent and B-cell-intrinsic mechanisms governed by PD-1 and PD-L1.

(A) Immunophenotyping of circulating Tfh (cTfh) cells among peripheral blood leukocytes from the PD-1-deficient patient (aged 10 or 11 years; three sampling time points), his healthy brother (aged 6 years), two PD-L1-deficient patients (aged 10 and 11 years), and healthy age-matched and adult controls.

(B) cTfh cell subsets.

(C) IL21 mRNA levels in PBMCs. TPM, transcript per million.

(D) IL-21 secretion by sorted naïve and memory CD4⁺ αβ T cells.

(E) IL-21 secretion by sorted memory CD4⁺ αβ T cells expanded for 2 weeks in vitro.

(F) IL-21 production by memory CD4⁺ T cells sorted from LNs or spleen of aged WT or Pdcd1^−/− mice (N=10 and 9, respectively) and expanded in vitro.

(G) Expression of PD-1 on blood B cells at baseline. The same data were previously published ¹⁸.

(H) Upregulation of PD-1 in activated B cells sorted from the PBMCs of healthy donors. Results compiled from three independent experiments.

(I and J) Expression of PD-L1 on blood B cells in PBMCs stimulated in vitro. (I) Representative plots. (J) Proportions of PD-L1⁺ cells in total B cells.

(K) Effect of IFN-γ neutralization on the induction of PD-L1 on B cells in PBMCs from healthy donors (N=6).

(L) Induction of PD-1, PD-L1, and PD-L2 on sorted naïve B cells from healthy donors (N=3~5) stimulated in vitro.

Graphs (A-F and H) depict the mean ± SEM. **** P < 0.0001; * P < 0.05; ns, not significant.

Expression of PD-1 and PD-L1 on human B cells

We hypothesized that loss of B-cell-autonomous PD-1:PD-L1-mediated signaling accounts for the impaired memory B-cell formation in PD-1 and PD-L1 deficiencies. To test this, we first analyzed expression of PD-1 on human B-cell subsets. Consistent with previous findings ⁶, PD-1 was detected in a minor population of resting blood B cells from healthy donors, but not the PD-1-deficient patient (Fig. 4G). Stimulation with CpG oligodeoxynucleotides, R848 (Resiquimod), or CD40L + IgM cross-linking upregulated PD-1 on naïve and memory B cells from healthy donors (Fig. 4H). By contrast, PD-1 expression was not detected on immature B cell subsets in BM from healthy donors (Fig. S4K). Moreover, we observed PD-L1 upregulation on B cells in anti-CD3/CD28-stimulated PBMCs from healthy controls, but not from PD-L1-deficient patients (Fig. 4I–J, S4L). IFN-γ blockade decreased PD-L1 induction on both naïve and memory B cells (Fig. 4K), suggesting IFN-γ secreted by activated T cells induces PD-L1 on B cells in a paracrine manner. We next investigated whether activated naïve B cells intrinsically upregulated PD-1 ligands. Stimulation for 48 hours with CD40L or anti-BCR F(ab’)₂ alone did not induce PD-L1. However, CpG alone modestly (~3-fold) induced PD-L1, and this was further increased (~6-fold) by the combination of CD40L + anti-BCR F(ab’)₂ + CpG (Fig. 4L). PD-1 or PD-L2 was not discernibly induced in these conditions (Fig. 4L). These findings support the hypothesis that PD-1:PD-L1 signals can affect B-cell differentiation or Ab responses in a B-cell-intrinsic manner upon activation.

Cell-intrinsic impairment of class switched Ig production by PD-1-deficient B cells

We investigated whether PD-1-deficient B cells have a cell-intrinsic impairment in Ig production in vitro (Fig. 5A). PD-1-deficient naïve B cells secreted less IgG than naïve B cells from healthy donors (Fig. 5B). IgA secretion by PD-1-deficient naïve B cells was also less than cells from adult controls, but within the range for age-matched controls, suggesting an age-dependent difference (Fig. 5B). IgM secretion by PD-1-deficient naïve B cells was also moderately lower than B cells from age-matched controls (Fig. 5B). The selective impairment in IgG secretion is consistent with the decrease in numbers of IgG⁺, but not IgA⁺, memory B cells in vivo (Fig. 1I). The impaired IgG secretion by PD-1-deficient B cells resembled that of STAT3 GOF naïve B cells, but was milder than naïve B cells deficient for c-Rel or RelB, or with monoallelic NFKB2 variants (Fig. 5B), consistent with mutations in NFKB2, RELB or CREL disrupting CD40 signaling in human B cells ^29–31. Likewise, PD-1-deficient naïve B cells generated fewer IgG⁺ cells de novo and specifically showed impaired secretion of IgG1 relative to other IgG subclasses or IgA (Fig. 5C–D, S5A). Furthermore, we found reduced cell counts following in vitro expansion of PD-1-deficient memory B cells compared to healthy controls, accompanied by impaired secretion of IgM and IgG1 (Fig. 5E, F). By contrast, secretion of IgA, IgG2, IgG3, and IgG4 was unaffected (Fig. 5F). Finally, anti-PD-1 blockade in vitro reduced the secretion of IgG and IgA, but not IgM, by CD40L + IL-21-stimulated naïve and memory B cells from healthy donors (Fig. 5G–I). These data indicate that human PD-1 plays a cell-intrinsic role in promoting the production of class switched Ig by naïve and memory B cells.

Figure 5. Cell-intrinsic impairment of antibody production in PD-1-deficient naïve and memory B cells.

(A) Schematic of the in vitro naïve B-cell stimulation assay.

(B) Ig secretion by naïve B cells from PD-1-deficient patient (aged 10 years; four technical replicates), healthy adults (N=13) and age-matched (N=2) donors.

(C) Two additional experiments similar to (B) for the PD-1-deficient patient (aged 11 years), his healthy brother (aged 6 years), and healthy adult (N=4) and age-matched (N=5) donors. Fold-changes in secreted Ig levels were calculated against the mean of healthy donors.

(D) flow cytometric analysis of surface Ig expression at the end of culture in (C).

(E-F) In vitro expansion of sorted memory B cells from the PD-1-deficient patient (aged 11 years), healthy brother (aged 6 years), and healthy age-matched (N=2) and adult (N=2) controls. Technical replicates were prepared for all controls and the PD-1-deficient patient (R=6). (E) Number of viable cells at day 14. (F) Secreted Ig levels.

(G-I) In vitro PD-1 blockade assay in naïve and memory B cells from healthy donors. (G) Functional validation of nivolumab and pembrolizumab biosimilars. Six technical replicates were prepared. Representative results from two independent experiments. (H) IgM and IgA secretion. (I) IgA and IgG secretion in the presence of mouse anti-human PD-1 neutralizing mAb.

Graphs depict mean ± SEM. ** P < 0.01; * P < 0.05; ns, not significant.

Impaired induction of c-Myc transcriptional signatures in PD-1-deficient naive B cells

We then explored the molecular basis of impaired IgG secretion due to B-cell-intrinsic PD-1 deficiency or blockade. To this end, we first performed bulk RNASeq on sorted PD-1-deficient naïve B cells stimulated with CD40L + IL-21 or anti-BCR F(ab’)₂ + CpG for 4, 24 and 48 hours (Fig. 6A). After 48 hours of CD40L + IL-21 stimulation, PD-1-deficient cells displayed impaired upregulation of IGHG1, IGHG3, and IGHG4, impaired downregulation of IGHD, but intact induction of IGHA1 and IGHA2 (Fig. S5B), providing further evidence of impaired class switching to IgG, but not IgA (Fig. 1I; Fig 5B, C, F). Weighted gene co-expression network analysis (WGCNA) ³² identified three co-expressed gene modules (M0, M1, and M2) differentially regulated between PD-1-deficient an control B cells (Fig. 6A). M1 and M2 correspond to early- and late-phase responses, induced at 4–24 and 24–48 hours, respectively, whereas M0 genes were expressed ex vivo but reduced following culture in vitro (Fig. 6A). M1 and M2 were enriched in c-Myc and E2F target genes and genes related to mTOR signaling (Fig. 6B). M2 was also enriched in genes involved in proliferation (mitotic spindle and G2/M checkpoints) (Fig. 6B). These results highlight cell-intrinsic defects in inducing transcriptional responses associated with c-Myc and proliferation following activation in PD-1-deficient naïve B cells.

Figure 6. Cell-intrinsic impairment of c-Myc expression and proliferation in PD-1-deficient naïve B cells.

(A-C) Bulk RNASeq of naïve B cells from the PD-1-deficient patient, his brother, IEI controls, and healthy adults and age-matched controls. (A) Three representative co-expressing gene modules differentially regulated in PD-1-deficient cells. (B) Top 3 Hallmark genesets overrepresented for each module. (C) Heatmap showing normalized mRNA levels for a subset of M2 genes upregulated in cells of age-matched controls following CD40L + IL-21 stimulation.

(D, E) CFSE assay in naïve B cells from the PD-1-deficient patient (aged 11 years), his brother (aged 6 years), multiple age-matched and adult controls. (D) CFSE dilution. (E) Proportions of CFSE^lo cells among total B cells or in Ig⁺ B-cell subsets.

(F, G) Proportions of total, naïve, and memory B cells co-expressing c-Myc and IRF4.

(H, I) Expression of c-Myc and IRF4 in naïve B cells cultured for 7 days. Graphs depict mean ± SEM.

CD40 signaling activates the canonical NF-κB pathway and c-Myc expression in mouse GC B cells ³³. Moreover, c-Rel-deficient murine GC B cells lack the c-Myc signature, and c-Rel deficiency or functional inactivation of c-Myc blunts established GCs in mice ^34,35. We hypothesized that the observed c-Myc-associated transcriptional responses in human naïve B cells are also driven by the canonical NF-κB pathway. We tested this by investigating the transcriptional responses of naïve B cells from patients with NFKB1 HI, dominant-negative (DN) RELA ³⁶, or STAT3 GOF and DN mutations, as IL-21 functions in a STAT3-dependent manner ³⁷. Following CD40L + IL-21 stimulation for 48 hours, NFKB1 HI naïve B cells displayed impaired induction of IGHG1, IGHG3, and IGHG4, similar to PD-1-deficient cells, whereas responses of STAT3 GOF, STAT3 DN, and RELA DN naïve B cells were similar to healthy donors (Fig. S5B). This is consistent with our previous findings of intact IgG class switching in naïve STAT3 DN B cells ³⁷. NFKB1 HI naïve B cells also displayed impaired induction of M1 and M2 gene modules following CD40L + IL-21 stimulation for 48 hours, whereas STAT3 GOF and STAT3 DN B cells resembled B cells from healthy donors (Fig. 6A). Moreover, among the 250 genes in M2 upregulated in naïve B cells from age-matched donors (including the healthy brother of the PD-1-deficient patient), 144 were hyper-induced, and 106 failed to be induced in RELA DN B cells (Fig. 6C). These findings suggest that 1) induction of c-Myc- and cell cycle-associated signatures and IgG class switching in human naïve B cells are governed predominantly by NF-κB p105/p50-dependent CD40 signals rather than IL-21/STAT3 signals, and 2) PD-1-deficient naïve B cells have intrinsic impairments in these pathways.

PD-1 and PD-L1 promote human B cell responses via c-Myc

The absence of cell cycle-associated transcriptional signatures in activated PD-1-deficient naïve B cells suggested impaired proliferation. Indeed, PD-1-deficient naïve B cells underwent less proliferation in response to CD40L + IL-21 (T-dependent) or anti-BCR F(ab’)₂ + CpG (T-independent) stimuli compared to control B cells (Fig. 6D). The impairment was particularly evident in PD-1-deficient IgG⁺ B cells upon anti-BCR F(ab’)₂ + CpG stimulation (Fig. 6E), likely reflecting impaired division-linked switching to IgG ³⁸. Given the reduced expression of c-Myc target genes in vivo and upon activation in vitro (Fig. 2E and 6A–B), and the indispensability of c-Myc for class switch recombination in murine B cells ³⁹, we next assessed c-Myc protein expression in PD-1-deficient B cells by flow cytometry ⁴⁰. Expression levels of c-Myc and IRF4 were higher in resting memory B cells than naïve B cells from healthy donors (Fig. 6F, G). Notably, the percentage of PD-1 deficient naïve B cells co-expressing c-Myc and IRF4 was less than healthy donor naïve B cells (Fig. 6F, G). We could not quantify c-Myc or IRF4 in PD-1-deficient memory B cells due to their scarcity (Fig. 6F). Stimulation of control naïve B cells with CD40L + IL-21 or anti-BCR F(ab’)₂ + CpG induced a distinct subset of c-Myc⁺IRF4⁺ cells, which was greatly reduced among PD-1-deficient naïve B cells (Fig. 6H, I). Unlike IRF4, Blimp1 was not correlated with c-Myc (Fig. S5C). These data demonstrate that PD-1 deficiency impairs c-Myc expression and proliferation of naïve B cells in a B-cell-intrinsic manner.

We explored common molecular mechanisms disrupted by PD-1 deficiency and in vitro PD-1 and PD-L1 blockade. PD-1 and PD-L1 blockade in vitro reduced expression of c-Myc protein and c-Myc-driven signature genes in CD40L + IL-21- or anti-BCR F(ab’)₂ + CpG-stimulated naïve and memory B cells from healthy donors (Fig. S5D–G). By contrast, we did not detect impairment in the proliferation or surface IgA and IgG expression following PD-1 or PD-L1 blockade in vitro (Fig. S5H–J). As PD-1 activates SHP2 ⁴¹, and SHP2 is required for the optimal c-Myc induction in tumor models, including GC-derived lymphoma ^42,43, we investigated the effect of pharmacological inhibition of SHP2 on the induction of c-Myc⁺IRF4⁺ cells from sort-purified naive B cells from healthy donors after in vitro stimulation. Indeed, the allosteric inhibitor of SHP2 RMC-4550 decreased the proportion of c-Myc⁺IRF4⁺ B cells induced by culture with CD40L + IL-21 or anti-BCR F(ab’)₂ + CpG (Fig. S6A). Moreover, while proliferation was largely unaffected (Fig. S6B), levels of IgG1 secreted by human memory B cells were consistently decreased by c-Myc (10058-F4, EN4, and KLPyr9) or SHP2 (RMC-4550 and SHP099) inhibitors (Fig. S6C). IgM levels were also decreased, but less markedly than IgG1, while IgA, IgG2, IgG3, and IgG4 were unaffected (Fig. S6C). These results suggest that B-cell-intrinsic PD-1:PD-L1 signaling via SHP2 promotes c-Myc expression, thus enhancing class switched Ig secretion.

Aberrant B-cell phenotypes in Pdcd1^fl/fl mb1-Cre mice

To definitively demonstrate the B cell-intrinsic role of PD-1 in humoral immunity, we generated mice with B cell-specific conditional deletion of PD-1 on the C57BL/6 background. Mice carrying a floxed version of Pdcd1 were crossed with mice carrying Cre expressed under the control of the Cd79a (mb1) promoter (referred to hereafter as Pdcd1^fl/fl mb1-Cre mice) (Fig. 7A). Cd79a encodes the Ig-α signaling subunit of the BCR and is expressed exclusively in B cells from the early pro-B cell stage ⁴⁴. Analysis of LPS-stimulated peripheral blood leukocytes revealed a clear induction of PD-1 on B cells from age-matched flox control (Pdcd1^fl/fl) mice but not on B cells from aged Pdcd1^fl/fl mb1-Cre mice (Fig. 7B, S7A). PD-1 expression on non-B cells was similar in Pdcd1^fl/fl and Pdcd1^fl/fl mb1-Cre mice (Fig. S7A, B). We then immunophenotypically analyzed Pdcd1^fl/fl mb1-Cre mice. Surprisingly, unlike constitutive Pdcd1^−/− mice, neither young (3–4 months old) nor aged (13–14 months old) Pdcd1^fl/fl mb1-Cre mice displayed splenomegaly (Fig. 7C); spleens from aged Pdcd1^fl/fl mb1-Cre mice were smaller (Fig. 7C) and LNs of aged Pdcd1^fl/fl mb1-Cre mice contained fewer cells (Fig. 7D) than age-matched flox control mice. A similar trend was observed for spleens of aged Pdcd1^fl/fl mb1-Cre mice, while the BM contained normal numbers of cells (Fig. 7D). By contrast, numbers of total and memory B cells were lower in the LN, spleen, and BM of aged Pdcd1^fl/fl mb1-Cre mice than age-matched flox control mice, with an enrichment of age-associated B cells (CD21/35⁻CD23⁻CD11c⁺T-bet⁺ cells) within the residual B cells (Fig. 7E–G; S7C–F). This phenotype contrasts the B-cell expansion in constitutive Pdcd1^−/− mice and the PD-1-deficient patient studied. Finally, deep phenotyping of BM of Pdcd1^fl/fl mb1-Cre mice identified a rare B-cell subset strongly expressing PD-1 that phenotypically resembles proliferating large pre-B cells (CD43⁺CD24⁺BP-1⁺Ki-67^hi) (Fig. 7H–J). These data demonstrate that the PD-1 expressed on B cells, probably during the large pre-B stage, is indispensable for B-cell homeostasis. Conversely, the phenotypic discordance between whole-body and B cell-specific deletion of PD-1 suggests that B cell-extrinsic PD-1-dependent mechanisms also contribute to the maintenance of B-cell compartments.

Figure 7. Altered immunophenotypes in mice with B-cell-specific deletion of PD-1.

(A) Schematic of cre-lox-mediated targeted disruption of Pdcd1 gene in B6 mice (Pdcd1^fl/fl mb1-Cre mice).

(B-D) Unimmunized young and aged Pdcd1^fl/fl mb1-Cre or flox control (Pdcd1^fl/fl) mice (N=6 per group). (B) PD-1 expression on LPS-stimulated blood B cells. (C) Spleen weight. (D) Absolute cell counts.

(E-J) B-cell phenotypes. N=6 for all groups; 2 experiments for BM were compiled. Absolute counts of (E) total and (F) memory B cells. (G) Proportions of age-associated B cells (ABCs).

(H-J) BM B cells. In (J), samples with too few cells were excluded from the analysis (PD-1, N=3 for Subset F from aged Pdcd1^fl/fl mb1-Cre mice; Ki-67, N=5 and N=4 for Subsets E and F from aged Pdcd1^fl/fl mb1-Cre mice, respectively). In (D-G and I), * P < 0.05; ** P < 0.01; **** P < 0.0001; ns, not significant.

Discussion

PD-1 is generally considered an inhibitory immune checkpoint. Our findings, however, support the concept originally proposed by Good-Jacobson et al. that PD-1 functions as a rheostat to restrain cytotoxic immunity while promoting humoral immunity by enabling the maturation of naïve B cells into high-affinity class switched memory B cells and PCs through B-cell-intrinsic and -extrinsic (i.e., Tfh-cell-dependent) mechanisms. Previous studies attributed defective humoral immunity in Pdcd1^−/− mice to impaired IL-21 production by Tfh cells ^7,9, while a B cell-intrinsic role of PD-1 was required for the development of long-lived PCs ⁷. By studying PD-1- and PD-L1-deficient patients, Pdcd1^−/−, Pdcd1lg2^−/−, Cd274^−/−Pdcd1lg2^−/−, and Pdcd1^fl/fl mb1-Cre mice, and in vitro cultured human B cells with PD-1 and PD-L1 blockade, we now provide compelling evidence that B cell-intrinsic PD-1 signaling promotes memory B cell formation and IgG secretion, increases BCR repertoire diversity and SHM, and shapes optimal Ab responses, thus governing B cell-mediated humoral immunity in mice and humans. These findings coherently link fragmentary findings in the literature, including poor reactivity of IgA from Pdcd1^−/− mice against commensal microbes ⁹, reduced PC numbers in Pdcd1^−/−, Cd274^−/−, Pdcd1lg2^−/−, or Cd274^−/−Pdcd1lg2^−/− mice ^7,8, and low-affinity influenza-specific Ab responses in cancer patients on anti-PD-1 immunotherapy ¹⁷. Overall, our findings extend knowledge of the indispensable roles of PD-1 in the development and function of Tfh cells and in the differentiation and function of naïve B cells to establish long-lived serological memory in a cell-intrinsic fashion in mice and humans.

Accumulating evidence suggests PD-1 blockade increases susceptibility to infections, predominantly bacterial pneumonia and lung abscesses, regardless of the use of corticosteroids or other immunosuppressive agents ^11–14. Fujita et al. reported infections in 19% (32/167) of patients with nonsmall cell lung carcinoma treated with the anti-PD-1 mAb nivolumab but without corticosteroids, with 78% of these infections caused by pyogenic bacteria (e.g., S. pneumoniae, H. influenzae), the control of which is Ab dependent ^13,15,16. Importantly, PD-1- and PD-L1-deficient patients had not yet manifested any severe bacterial infections typically occurring in patients with Ab deficiency, suggesting that the observed defects in humoral immunity are probably insufficient to cause clinically overt phenotypes during childhood. Nevertheless, most cancer patients on PD-1 or PD-L1 blockade immunotherapy are already immunocompromised due to their malignancy, comorbidity, and older age. Thus, subtle perturbations of humoral immunity due to PD-1 or PD-L1 blockade can potentially trigger devastating bacterial infections. These findings should prompt further investigations into the frequency and severity of bacterial and other infections, and the potential clinical benefits of prophylactic vaccination (e.g., S. pneumoniae and H. influenzae) and low-dose intravenous Ig in cancer patients on PD-1 or PD-L1 blockade immunotherapy.

Limitations of the study

One limitation of our study is the small number of patients with inherited PD-1 or PD-L1 deficiency. Although expected to be rare, future discoveries of new patients with related genotypes would confirm and broaden the insights described here. Another limitation is that we have not fully clarified the relative contributions of PD-1:PD-L1-mediated B-cell-extrinsic (governed by Tfh cells and possibly other cell types) and B-cell-intrinsic mechanisms in shaping various aspects of B-cell memory and Ab immunity. Given the large phenotypic discordance between constitutive vs. conditional PD-1 deletion in mice at baseline, further investigations in Ag-exposed settings, such as different types of vaccines, are necessary to fully elucidate the significance of PD-1 on B cells vs. other leukocytes in shaping optimal humoral immunity.

STAR Methods

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Stuart G. Tangye (s.tangye@garvan.org.au).

Materials availability

Pdcd1 floxed C57BL/6 mouse line has been deposited to the RIKEN Center for Developmental Biology (CDB) [Accession: CDB1220K].

Data and Code availability

• Bulk RNA-seq data have been deposited at NCBI Sequence Read Archive and are publicly available as of the date of publication. Accession numbers are listed in the key resources table.

• Processed single-cell and bulk RNA-seq datasets and accompanying original codes have been deposited at Mendeley Data and are publicly available as of the date of publication. DOIs are listed in the key resources table.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Key resources table.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
V450 anti-human CD3 (Clone: UCHT1)	BD Biosciences	Cat# 560365
BUV563 anti-human CD4 (Clone: SK3)	BD Biosciences	Cat# 612913
BUV737 anti-human CD8 (Clone: SK1)	BD Biosciences	Cat# 612754
BUV395 anti-human CD14 (Clone: Mﾏ · P9)	BD Biosciences	Cat# 563561
PE/Dazzle 594 anti-human CD16 (Clone: 3G8)	BioLegend	Cat# 302053
BV785 anti-human CD20 (Clone: 2H7)	BioLegend	Cat# 302356
BV605 anti-human CD56 (Clone: 5.1H11)	BioLegend	Cat# 362537
Alexa Fluor 647 anti-human αβTCR (Clone: B1)	BioLegend	Cat# 331214
FITC anti-human Vδ1 (Clone: REA173)	Miltenyi Biotec	Cat# 130-118-498
APC/Fire 750 anti-human Vδ2 (Clone: B6)	BioLegend	Cat# 331419
Alexa Fluor 700 anti-human Vα7.2 (Clone: 3C10)	BioLegend	Cat# 351727
BV480 anti-human iNKT (Clone: 6B11)	BD Biosciences	Cat# 746788
APC anti-human Vβ11 (Clone: REA559)	Miltenyi Biotec	Cat# 130-125-508
BUV395 anti-human CD11b (Clone: ICRF44)	BD Biosciences	Cat# 563840
BUV496 anti-human CD21 (Clone: B-ly4)	BD Biosciences	Cat# 750614
BUV563 anti-human CD24 (Clone: ML5)	BD Biosciences	Cat# 741364
BUV615 anti-human CD138 (Clone: MI15)	BD Biosciences	Cat# 751148
BUV661 anti-human CD38 (Clone: HIT2)	BD Biosciences	Cat# 612969
BUV737 anti-human CD10 (Clone: HI10a)	BD Biosciences	Cat# 741825
BUV805 anti-human CD23 (Clone: EBVCS-5)	BD Biosciences	Cat# 749448
BV421 anti-human CXCR4 (Clone: 12G5)	BioLegend	Cat# 306518
V450 anti-human CD56 (Clone: B159)	BD Biosciences	Cat# 560361
BV480 anti-human CD5 (Clone: UCHT2)	BD Biosciences	Cat# 566122
VioGreen anti-human IgA (Clone: IS11-8E10)	Miltenyi Biotec	Cat# 130-113-481
BV570 anti-human IgM (Clone: MHM-88)	BioLegend	Cat# 314517
BV605 anti-human PD-L1 (Clone: MIH1)	BD Biosciences	Cat# 740426
BV650 anti-human CD19 (Clone: HIB19)	BD Biosciences	Cat# 740568
BV711 anti-human PD-L2 (Clone: MIH18)	BD Biosciences	Cat# 564258
BV750 anti-human B220 (Clone: RA3-6B2)	BioLegend	Cat# 103261
BV785 anti-human IgD (Clone: IA6-2)	BioLegend	Cat# 348242
BB515 anti-human CD40 (Clone: 5C3)	BD Biosciences	Cat# 565258
FITC anti-human CD80 (Clone: 2D10)	BioLegend	Cat# 305206
Alexa Fluor 532 anti-human CD20 (Clone: 2H7)	eBioscience	Cat# 58-0209-42
BB700 anti-human PD-1 (Clone: EH12.1)	BD Biosciences	Cat# 566461
PE/Cyanine5 anti-human IgG (Clone: G18-145)	BD Biosciences	Cat# 551497
APC anti-human CD27 (Clone: M-T271)	BD Biosciences	Cat# 561400
Alexa Fluor 647 anti-human CD86 (Clone: IT2.2)	BioLegend	Cat# 305416
Alexa Fluor 700 anti-human CD11c (Clone: Bu15)	BioLegend	Cat# 337220
APC/Fire 750 anti-human BCMA (Clone: 19F2)	BioLegend	Cat# 357516
APC/Fire 810 anti-human HLA-DR (Clone: L243)	BioLegend	Cat# 307674
PE anti-human IRF4 (Clone: IRF4.3E4)	BioLegend	Cat# 646404
PE/CF594 anti-human Blimp1 (Clone: 6D3)	BD Biosciences	Cat# 565274
PE/Cyanine7 anti-human T-bet (Clone: 4B10)	BioLegend	Cat# 644824
PerCP/eFluor 710 anti-human RORγ (Clone: AFKJS-9)	eBioscience	Cat# 46-6988-82
PE/Cyanine7 anti-human αβTCR (Clone: IP26)	BioLegend	Cat# 306720
FITC anti-human CD3 (Clone: UCHT1)	Cytek Biosciences	Cat# SKU 35-0038-T500
PE anti-human CD4 (Clone: OKT4)	BioLegend	Cat# 317410
BV711 anti-human CD8 (Clone: RPA-T8)	BioLegend	Cat# 301044
Alexa Fluor 700 anti-human CD20 (Clone: 2H7)	BD Biosciences	Cat# 560631
BV421 anti-human CD38 (Clone: HIT2)	BioLegend	Cat# 303526
BV510 anti-human CD45RA (Clone: HI100)	BD Biosciences	Cat# 563031
BUV395 anti-human CCR7 (Clone: 2-L1-A)	BD Biosciences	Cat# 563977
BUV661 anti-human CD138 (Clone: MI15)	BD Biosciences	Cat# 749873
BV421 anti-human CD8 (Clone: RPA-T8)	BioLegend	Cat# 562428
BV480 anti-human CD123 (Clone: 9F5)	BD Biosciences	Cat# 566182
BV711 anti-human PD-L1 (Clone: 29E.2A3)	BioLegend	Cat# 329722
PE/Cyanine7 anti-human CD20 (Clone: 2H7)	BioLegend	Cat# 302312
Spark NIR 685 anti-human CD14 (Clone: 63D3)	BioLegend	Cat# 367150
Alexa Fluor 647 anti-human c-Myc (Clone: C-33)	Santa Cruz Biotechnology	Cat# sc-42 Af647
PE anti-human CD5 (Clone: L17F12)	BD Biosciences	Cat# 347307
APC anti-human CD10 (Clone: HI10a)	BD Biosciences	Cat# 340923
BV711 anti-human CD19 (Clone: SJ25C1)	BD Biosciences	Cat# 563036
Pacific Blue anti-human CD20 (Clone: 2H7)	BioLegend	Cat# 302328
BUV563 anti-human CD21 (Clone: B-ly4)	BD Biosciences	Cat# 741362
BV421 anti-human CD21 (Clone: B-ly4)	BD Biosciences	Cat# 562966
FITC anti-human CD23 (Clone: Tü1)	Thermo Fisher	Cat# MHCD2301
PE anti-human CD27 (Clone: M-T271)	BD Biosciences	Cat# 555441
BB515 anti-human CD27 (Clone: M-T271)	BD Biosciences	Cat# 564642
APC anti-human CD38 (Clone: HB7)	BD Biosciences	Cat# 340439
APC anti-human IgG (Clone: G18-145)	BD Biosciences	Cat# 550931
PE/Cyanine5 anti-human IgA (Clone: G20-359)	BD Biosciences	Custom Ab
APC anti-human CD4 (Clone: RPA-T4)	BioLegend	Cat# 300514
BUV395 anti-human CD45RA (Clone: HI100)	BD Biosciences	Cat# 568712
BB700 anti-human CD127 (Clone: HIL-7R-M21)	BD Biosciences	Cat# 566398
BYG584 anti-human CD25 (Clone: 2A3)	BD Biosciences	Custom Ab
BUV615 anti-human CXCR5 (Clone: RF8B2)	BD Biosciences	Cat# 751293
BV421 anti-human CXCR3 (Clone: G025H7)	BioLegend	Cat# 353716
PE/CF594 anti-human CCR6 (Clone: 11A9)	BD Biosciences	Cat# 564816
BUV661 anti-human PD-1 (Clone: EH12.1)	BD Biosciences	Cat# 750260
BV605 anti-human PD-1 (Clone: EH12.1)	BD Biosciences	Cat# 563245
PE anti-human PD-L1 (Clone: 29E.2A3)	BioLegend	Cat# 329706
APC anti-human PD-L2 (Clone: 24F.10C12)	BioLegend	Cat# 329608
FITC anti-human CD25 (Clone: 2A3)	BD Biosciences	Cat# 347643
PerCP/Cyanine5.5 anti-human CD45RA (Clone: HI100)	Thermo Fisher	Cat# 45-0458-42
Alexa Fluor 647 anti-human CXCR5 (Clone: RF8B2)	BD Biosciences	Cat# 558113
PE anti-human CCR6 (Clone: 11A9)	BD Biosciences	Cat# 559562
BUV737 anti-human CD4 (Clone: SK3)	BD Biosciences	Cat# 612748
BUV395 anti-human CD20 (Clone: 2H7)	BD Biosciences	Cat# 563782
BUV737 anti-human CD10 (Clone: H10a)	BD Biosciences	Cat# 612826
PE/Cyanine7 anti-human CD27 (Clone: M-T271)	BD Biosciences	Cat# 560609
BV605 anti-human IgG (Clone: G18-145)	BD Biosciences	Cat# 563246
Biotin anti-human IgA (Polyclonal)	Southern Biotech	Cat# 2052-08
BV570 anti-human CD3 (Clone: UCHT1)	BioLegend	Cat# 300436
BUV805 anti-human CD20 (Clone: 2H7)	BD Biosciences	Cat# 612905
BV605 anti-human iNKT (Clone: 6B11)	BioLegend	Cat# 342930
BV711 anti-human γδTCR (Clone: 11F2)	BD Biosciences	Cat# 745505
BUV496 anti-human CD8 (Clone: RPA-T8)	BD Biosciences	Cat# 612942
PE/Cyanine7 anti-human CCR7 (Clone: G043H7)	BioLegend	Cat# 353226
APC/Cyanine7 anti-human CD3 (Clone: SK7)	Cytek Biosciences	Cat# SKU 25-0036-T100
BUV395 anti-human CD4 (Clone: SK3)	BD Biosciences	Cat# 563552
FITC anti-human CD25 (Clone: BC96)	BioLegend	Cat# 302604
APC anti-human CD127 (Clone: A019D5)	BioLegend	Cat# 351342
BV786 anti-human CD45RA (Clone: HI100)	BD Biosciences	Cat# 563870
Pacific Blue anti-human CCR7 (Clone: G043H7)	BioLegend	Cat# 353210
PE/Cyanine7 anti-human CXCR5 (Clone: J252D4)	BioLegend	Cat# 356923
AmCyan anti-human CXCR3 (Clone: 1C6)	BD Biosciences	Custom Ab
BUV737 anti-human CCR6 (Clone: 11A9)	BD Biosciences	Cat# 612780
PerCP/Cyanine5.5 anti-human CD10 (Clone: HI10a)	BD Biosciences	Cat# 563508
APC/Alexa Fluor700 anti-human CD19 (Clone: J3-119)	Beckman Coulter	Cat# B49212
PerCP/Cyanine5.5 anti-human CD19 (Clone: HIB19)	BD Biosciences	Cat# 561295
FITC anti-human CD21 (Clone: B-ly4)	BD Biosciences	Cat# 561372
PE anti-human CD27 (Clone: M-T271)	BD Biosciences	Cat# 560985
PC7 anti-human CD34 (Clone: 581)	BD Biosciences	Cat# 560710
FITC anti-human CD86 (Clone: 2331 (FUN-1))	BD Biosciences	Cat# 557343
PE/CF594 anti-human IgM (Clone: G20-127)	BD Biosciences	Cat# 562539
Alexa Fluor 647 anti-human PD-1 (Clone: MIH4)	BD Biosciences	Cat# 566851
Alexa Fluor 647 mouse IgG1 Isotype control (Clone: MOPC-21)	BD Biosciences	Cat# 557732
APC anti-human CD3 (Clone: UCHT1)	Cytek Biosciences	Cat# 20-0038-T500
PE/Dazzle 594 anti-human IFN-γ (Clone: 4S.B3)	BioLegend	Cat# 502546
BV711 anti-human TNF (Clone: MAb11)	BioLegend	Cat# 502940
BUV395 anti-mouse CD19 (Clone: 1D3)	BD Biosciences	Cat# 563557
BV786 anti-mouse CD45.2 (Clone: 104)	BD Biosciences	Cat# 563686
BUV737 anti-mouse CD45.2 (Clone: 104)	BD Biosciences	Cat# 612778
PE anti-mouse/human CD45R (B220) (Clone: RA3-6B2)	eBioscience	Cat# 12-0452-8
FITC anti-mouse/human CD45R (B220) (Clone: RA3-6B2)	Cytek Biosciences	Cat# 35-0452
BV785 anti-mouse/human CD45R (B220) (Clone: RA3-6B2)	BioLegend	Cat# 103246
BV510 anti-mouse CD80 (Clone: 16-10A1)	BD Biosciences	Cat# 740130
BUV737 anti-mouse CD80 (Clone: 16-10A1)	BD Biosciences	Cat# 612773
PE anti-mouse CD73 (Clone: TY/11.8)	BioLegend	Cat# 127206
PerCP/Cyanine5.5 anti-mouse CD73 (Clone: TY/11.8)	BioLegend	Cat# 127213
APC/Cyanine7 anti-mouse CD3 (Clone: 17A2)	BioLegend	Cat# 100222
APC/Cyanine7 anti-mouse F4/80 (Clone: BM8)	BioLegend	Cat# 123118
PE Anti-mouse CD43 (Clone: S7)	BD Biosciences	Cat# 553271
PE/Cyanine7 anti-mouse IgM (Clone: II/41)	eBioscience	Cat# 25-5790-82
FITC anti-mouse Ly-51 (CD249) (Clone: 6C3)	BioLegend	Cat# 108305
Pacific Blue anti-mouse IgD (Clone: 11-26c.2a)	BioLegend	Cat# 405712
BV510 anti-mouse CD24 (Clone: M1/69)	BioLegend	Cat# 101831
APC anti-mouse PD-1 (Clone: RMP1-30)	BioLegend	Cat# 109112
PE/Cyanine7 anti-mouse PD-1 (Clone: RMP1-30)	BioLegend	Cat# 109110
BUV395 anti-mouse PD-1 (Clone: 29F.1A12)	BD Biosciences	Cat# 568596
PE/Cyanine7 anti-mouse CD23 (Clone: B3B4)	BioLegend	Cat# 101613
FITC anti-mouse CD21/CD35 (CR2/CR1) (Clone: 7E9)	BioLegend	Cat# 123407
BV421 anti-mouse CD11c (Clone: HL3)	BD Biosciences	Cat# 562782
APC anti-mouse CD95 (Clone: 15A7)	eBioscience	Cat# 17-0951-82
PE anti-mouse CD83 (Clone: Michel-19)	BioLegend	Cat# 121508
PE/Cyanine7 anti-mouse CD38 (Clone: 90)	BioLegend	Cat# 102718
BV421 anti-mouse CD184 (CXCR4) (Clone: L276F12)	BioLegend	Cat# 146511
BV510 anti-mouse CD86 (Clone: GL1)	BD Biosciences	Cat# 563077
APC mouse IgG1 Isotype Control (Clone: MOPC-21)	Cytek Biosciences	Cat# 20-4714
PE rat IgG1 Isotype Control (Clone: 1H5)	MBL International	Cat# M080-5
PE/Cyanine7 rat IgG2a, κ Isotype Ctrl (Clone: RTK2758)	BioLegend	Cat# 400522
APC rat IgG2b, κ Isotype Ctrl (Clone: RTK4530)	BioLegend	Cat# 400611
APC anti-mouse/human T-bet (Clone: 4B10)	BioLegend	Cat# 644813
APC anti-mouse/human T-bet (Clone: REA102)	Miltenyi Biotec	Cat# 130-119-783
BUV395 anti-mouse Ki-67 (Clone: B56)	BD Biosciences	Cat# 564071
FITC anti-mouse/human c-Myc (Clone: C-33)	Santa Cruz Biotechnology	Cat# sc-42 FITC
APC human IgG1 Isotype control (Clone: REA293)	Miltenyi Biotec	Cat# 130-120-709
FITC mouse IgG1 Isotype control (Clone: MOPC-21)	Cytek Biosciences	Cat# 35-4714
APC anti-mouse IgA (Clone: mA-6E1)	eBioscience	Cat# 17-4204-82
APC anti-mouse CD3 (Clone: 17A2)	BioLegend	Cat# 100236
FITC anti-mouse CD19 (Clone: 1D3)	Cytek Biosciences	Cat# 35-0193
APC anti-mouse/human CD44 (Clone:IM7)	BioLegend	Cat# 103011
PE/Cyanine7 anti-mouse CD4 (Clone:RM4-5)	eBioscience	Cat# 25-0042-82
BV510 anti-mouse CD45RB (Clone: 16A)	BD Biosciences	Cat# 740107
FITC anti-mouse CD44 (Clone: IM7)	BD Bioscience	Cat# 561859
BV786 anti-mouse CD4 (Clone: H129.19)	BD Biosciences	Cat# 740873
BUV395 anti-mouse CD45.2 (Clone: 104)	BD Biosciences	Cat# 564616
PE anti-mouse IL-21 (Clone: mhalx21)	eBioscience	Cat# 12-7213-82
Anti-human CD16 (Clone: B73.1)	eBioscience	Cat# 16-0167-82
Anti-human CD32 (Clone: 6C4)	eBioscience	Cat# 16-0329-81
Anti-mouse CD16/CD32 (Clone: 2.4G2)	BD Biosciences	Cat# 553142
Anti-human CD3/CD19 bispecific molecule	BPS Bioscience	Cat# 100441
Anti-human PD-1-hIgG4 antibody (S228P) (Nivolumab biosimilar)	InvivoGen	Cat# hpd1ni-mab114
Anti-human PD-1-hIgG4 antibody (S228P) (Pembrolizumab biosimilar)	InvivoGen	Cat# hpd1pe-mab14
Anti-β-Gal-hIgG4 (S228P)	InvivoGen	Cat# bgal-mab114
Anti-human PD-L1-hIgG1 antibody (N298A) (Atezolizumab biosimilar)	InvivoGen	Cat# hpdl1-mab12
Anti-β-Gal-hIgG1 (N298A)	InvivoGen	Cat# bgal-mab12
Anti-human PD-1 moues IgG1,κ (Clone: PD1.D3)	GenScript	Cat# A01829
Anti-human IFN-γ moues IgG1,κ (Clone: NIB42)	eBioscience	Cat# 16-7318-81
Mouse IgG1 Isotype Control (MOPC-21)	BioLegend	Cat# 400166
AffiniPure™ F(ab’)2 fragment goat anti-human IgM, FC5μ fragment specific (Polyclonal)	Jackson Immunoresearch	Cat# 109-006-129
AffiniPure™ F(ab’)2 fragment goat anti-human IgA + IgG + IgM (H+L)	Jackson Immunoresearch	Cat# 109-006-064
Anti-HA mAb	R&D Systems	Cat# MAB060
Bacterial and virus strains
Biological samples
Privigen human IVIg product	CSL Behring	N/A
Immunoglobulin G (IgG) depleted human serum	Molecular Innovations, Inc	Cat# HPLASERGFA5ML
Chemicals, peptides, and recombinant proteins
BD Horizon Brilliant Stain Buffer Plus	BD Biosciences	Cat# 566385
FcR Blocking Reagent, human	Miltenyi Biotec	Cat# 130-059-901
BV421 MR1 tetramer	NIH Tetramer Core	N/A
Zombie NIR Fixable Viability dye	BioLegend	Cat# 423106
Ghost Dye Red 780	Cytek Biosciences	Cat# SKU 13-0865-T100
DAPI (4’,6-diamidino-2-phénylindole, dichlorhydrate)	Sigma-Aldrich	Cat# MBD0015
7-amino-actinomycin D (7-AAD)	eBioscience	Cat# 00-6993-50
CFSE	Cytek Biosciences	Cat# SKU 13-0850-U500
Brefeldin A	eBioscience	Cat# 00-4506-51
Brefeldin A	Cytek Biosciences	Cat# SKU TNB-4506-L001
Monensin	Cytek Biosciences	Cat# SKU TNB-4505-L001
Foxp3 / Transcription Factor Staining Buffer Set	eBioscience	Cat# 00-5523-00
Foxp3 / Transcription Factor Staining Buffer Kit	Cytek Biosciences	Cat# SKU TNB-0607-KIT
ImmunoCult™ Human CD3/CD28/CD2 T-Cell Activator	STEMCELL Technologies	Cat# 10970
Cell Stimulation Cocktail	eBioscience	Cat# 00-4970-93
ImmunoCult™-XF T Cell Expansion Medium	STEMCELL Technologies	Cat# 10981
Recombinant human IL-2	Roche Diagnostics Corporation	Cat# 11147528001
CD4(L3T4) MicroBeads, mouse	Miltenyi Biotec	Cat# 130-117-043
Dynabeads™ Mouse T-Activator CD3/CD28 for T-Cell Expansion and Activation	Gibco	Cat# 11453D
Phorbol 12-myristate 13-acetate (PMA)	Sigma-Aldrich	Cat# P8139
Ionomycin Calcium Salt	Sigma-Aldrich	Cat# I0634
LPS	Sigma-Aldrich	Cat# L7261
Heat Killed E.coli 0111:B4 (HKEB)	InvivoGen	Cat# tlrl-hkeb2
Heat-killed Mycobacterium tuberculosis (HKMT)	InvivoGen	Cat# tlrl-hkmt-1
Beta-1,3-glucan from Alcaligenes faecalis (Curdlan)	InvivoGen	Cat# tlrl-curd
Recombinant Human CD40 Ligand	InVivoGen	Cat# rcyec-hcd40l
Recombinant Human CD40 Ligand, HA-tagged	R&D Systems	Cat# 6420-CLB
MEGACD40L Protein (soluble) (human)	Enzo Life Sciences	Cat# ALX-522-110-C010
Recombinant human IL-21	PeproTech	Cat# 200-21
CpG ODN 2006 (ODN 7909)	InvivoGen	Cat# tlrl-2006
R848 (Resiquimod), Imidazoquinoline compound	InvivoGen	Cat# tlrl-r848-1
SHP099	MedChem Express	Cat# HY-100388
ImmunoCult™ Human B Cell Expansion Kit	STEMCELL Technologies	Cat# 100-0645
RMC-4550	MedChem Express	Cat# HY-116009
10058-F4	MedChem Express	Cat# HY-12702
EN4	MedChem Express	Cat# HY-134761
KJ Pyr 9	MedChem Express	Cat# HY-19735
Critical commercial assays
LEGENDplex Human Immunoglobulin Isotyping assay	BioLegend	Cat# 740638
Human IL-21 ProQuantum Immunoassay Kit	Thermo Fisher	Cat# A35593
RNeasy Plus Micro Kit	Qiagen	Cat# 74034
SMART-Seq v4 Ultra Low Input RNA Kit	Clontech	Cat# 634888
Nextera XT DNA Library Preparation Kit	Illumina	Cat# FC-131-1024
Deposited data
Bulk RNA sequencing data for sorted naïve and memory B cells of the PD-1-deficient patient and his healthy brother, at baseline or with stimulation in vitro	This paper	NCBI Sequence Read Archive (accession no. PRJNA1141130)
Bulk RNA sequencing data for unstimulated whole-blood leukocytes of the PD-L1-deficient patients and age-matched controls	Johnson, Ogishi, and Domingo-Vila et al. 23	NCBI Sequence Read Archive (accession no. PRJNA1084900)
Single-cell RNA sequencing data for the PD-1-deficient patient and his healthy brother	Ogishi et al. 47	NCBI Sequence Read Archive (accession no. PRJNA723618); Mendeley Data (https://doi.org/10.17632/nb26v3mx3x.2)
Single-cell RNA sequencing data for the PD-L1-deficient patients and an age-matched control	Johnson, Ogishi, and Domingo-Vila et al. 23	NCBI Sequence Read Archive (accession no. PRJNA1084900); Mendeley Data (https://doi.org/10.17632/4fwbcswj8d.1)
Bulk RNA sequencing data for B cell subsets before and after influenza vaccination in patients on PD-1 blockade immunotherapy	Herati et al. 18	GEO: GSE179487
Bulk RNA sequencing data for WT and Pdcd1−/− mice	This paper	NCBI Sequence Read Archive (accession no. PRJNA1141130)
Bulk RNA sequencing data for in vitro stimulated PBMCs of the PD-1-deficient patient and his healthy brother	This paper	NCBI Sequence Read Archive (accession no. PRJNA1141130)
Bulk RNA sequencing data for in vitro stimulated PBMCs of the PD-L1-deficient patients and an age-matched control	Johnson, Ogishi, and Domingo-Vila et al. 23	NCBI Sequence Read Archive (accession no. PRJNA1084900)
Bulk RNA sequencing data for sorted naïve B cells from healthy donors with in vitro PD-1 or PD-L1 blockade	This paper	NCBI Sequence Read Archive (accession no. PRJNA1141130)
Human reference genome NCBI build 37, GRCh37.p13	Genome Reference Consortium	https://www.ncbi.nlm.nih.gov/grc/human
Mouse reference genome NCBI build 38, GRCm38.p6	Genome Reference Consortium	https://www.ncbi.nlm.nih.gov/grc/mouse
MsigDB Human Collections	MsigDB Molecular Signatures Database	https://www.gsea-msigdb.org/gsea/msigdb
Experimental models: Cell lines
HuT78	ATCC	TIB-161
Raji	ATCC	CCL-86
HK3i embryonic stem cells	Kiyonari et al. 30	N/A
Experimental models: Organisms/strains
Mouse: Pdcd1 floxed (Pdcd1fl/fl) C57BL/6	This paper	CDB Acc.No.: CDB1220K
Mouse: CAG-FLPe deleter C57BL/6	Kanki et al. 25	N/A
Mouse: Cd79a (mb1) Cre/+ C57BL/6	Hobeika et al. 19	N/A
Mouse: C57BL/6NJcl	CLEA Japan	RRID:IMSR_JCLJCL:MIN-0004
Mouse: C57BL/6NCrSlc	Japan SLC	RRID:MGI:5295404
Mouse: Pdcd1−/− C57BL/6	Nishimura et al. 45	RBRC No.: RBRC02142
Oligonucleotides
Primers: Pdcd1 genotyping Fwd GCTGCTGGCTTTCCTAAACA	Thermo Fisher Scientific	N/A
Primers: Pdcd1 genotyping Rev TGTCAGGCACTGAAGAGATCTACAC	Thermo Fisher Scientific	N/A
Recombinant DNA
Software and algorithms
FlowJo v10	FlowJo, LLC	https://www.flowjo.com/
Kaluza Analysis Software v2.2.1	Beckman Coulter	https://www.beckman.com/flow-cytometry/software/kaluza/downloads
R v4.4.0	The R Project for Statistical Computing	https://www.r-project.org/
Rstudio Desktop	Posit	https://posit.co/download/rstudio-desktop/
uwot v0.2.2	CRAN	https://cran.r-project.org/web/packages/uwot/index.html
Monocle3 v1.3.1	Cao et al. 7	https://cole-trapnell-lab.github.io/monocle3/docs/installation/
fgsea v1.30.0	Bioconductor	https://bioconductor.org/packages/release/bioc/html/fgsea.html
fastqc v0.12.1	Babraham Bioinformatics	https://www.bioinformatics.babraham.ac.uk/projects/fastqc/
STAR aligner v2.6	Dobin et al. 10	https://github.com/alexdobin/STAR
featureCounts v1.6.0	Liao et al. 35	http://subread.sourceforge.net
DESeq2 v1.44.0	Love et al. 37	https://bioconductor.org/packages/release/bioc/html/DESeq2.html
decoupleR v2.10.0	Badia-i-Mompel et al. 3	https://www.bioconductor.org/packages/release/bioc/html/decoupleR.html
MiXCR v4.3.2	Bolotin et al. 5	https://mixcr.com/
alakazam v1.3.0	Gupta et al. 15	https://cran.r-project.org/web/packages/alakazam/index.html
iNEXT v3.0.1	Hsieh et al. 21	https://johnsonhsieh.github.io/iNEXT/
WGCNA v1.72-5	Langfelder & Horvath 31	https://cran.r-project.org/web/packages/WGCNA/index.html
hypeR v2.2.0	Federico & Monti 11	https://www.bioconductor.org/packages/release/bioc/html/hypeR.html
limma v3.60.3	Law et al. 32	https://bioconductor.org/packages/release/bioc/html/limma.html
Code for single-cell trajectory analysis, BCR repertoire analysis, and bulk RNA sequencing analysis	This paper, Mendeley Data	Mendeley Data (DOI: 10.17632/398bg5r38x.1)
Other

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Human subjects

Healthy volunteers were recruited at Rockefeller University and the Garvan Institute of Medical Research. Buffy coats collected from healthy blood donors were purchased from the Australian Red Cross Blood Service. The PD-1-deficient patient and his healthy brother and parents ¹⁸ were recruited at the Necker Hospital for Sick Children. The PD-L1-deficient patients and their healthy mother ¹⁹ were recruited at the Genetic Beta-Cell research bank at the University of Exeter, UK, for genetic testing to determine the cause of their infancy-onset diabetes. Healthy controls matched with the PD-L1-deficient patients for age were recruited at the Benaroya Research Institute in the United States. Patients heterozygous for STAT3 gain-of-function (GOF) mutations (R152W, M394T, F408L, G419R, T663I, and T716M) were recruited at the Necker Hospital for Sick Children, the National Institute of Allergy and Infectious Diseases, and the Garvan Institute of Medical Research. A patient heterozygous for a STAT3 dominant-negative (DN) mutation (E382Q) was recruited at Hiroshima University. A patient heterozygous for an NFKB1 loss-of-function (LOF) mutation (E199X) was recruited at Rockefeller University. Patients heterozygous for an NFKB2 LOF mutation (A867V) were recruited at the Garvan Institute of Medical Research. Patients heterozygous for RELA DN mutations [Q389*, E473Rfs*18, and c.1034–1G>A (an essential splice site mutation)] were recruited at the universities of Tohoku, Kyoto, and Kyushu ³⁶. Written informed consent was obtained from all patients, relatives, and healthy volunteers enrolled in this study. The study was approved by the institutional ethics committees of Rockefeller University, Necker Hospital for Sick Children, University of Exeter, King’s Colledge London, Benaroya Research Institute, and Garvan Institute of Medical Research, and was performed in accordance with the requirements of these bodies. Experiments on samples from human subjects were conducted in the United States, France, United Kingdom, and Australia, in accordance with local regulations and with the approval of the institutional review board of the corresponding institution.

Mice

For immunization experiments, Pdcd1^−/− C57BL/6 mice (RBRC02142) ⁴, Pdcd1lg2^−/− C57BL/6 mice ²⁵, or Cd274^−/−Pdcd1lg2^−/− BALB/c mice ²⁶ were used, together with corresponding wild-type (WT) mice. These animal experiments were approved by the Yale Institutional Animal Care and Use Committee. For immunophenotyping and bulk RNASeq experiments, WT or Pdcd1^−/− C57BL/6 mice were used. These animal experiments were performed according to guidelines approved by the Kyoto University institutional review board. Experiments related to the construction of a Pdcd1^fl/fl mb1-Cre mouse line and their immunophenotyping analysis were approved by the Institutional Animal Care and Use Committee of the RIKEN Kobe Branch and the Kyoto University institutional review board.

METHOD DETAILS

Analysis of Ig isotypes in plasma samples

Immunoglobulin isotypes in the plasma samples of the PD-1-deficient patient, his relatives (father, mother, and brother), travel controls, and local controls were quantified with the LEGENDplex Human Immunoglobulin Isotyping Panel kit (BioLegend, 740638). The patient and his father were sampled twice at different time points. Samples were diluted 1:5000 before measurements.

Phage immunoprecipitation sequencing for microbial antigens (VirScan)

Phage immunoprecipitation sequencing (PhIP-Seq) was performed in a previous study ¹⁸. Briefly, serum samples were obtained from the patient, his healthy six-year-old brother, and his parents at two sampling points (at the initial presentation and during follow-up). Serum samples from five age-matched controls and 13 patients with biallelic CARMIL2 mutations suffering from combined immunodeficiency ^20,21 were also included in the analysis. Moreover, Privigen^® (CSL Behring), a concentrated human IgG product derived from pooled human plasma for intravenous immunoglobulin (IVIg) therapy, and IgG-depleted serum (Molecular Innovations, Inc.; Cat: HPLASERGFA5ML) were used as the positive and negative controls, respectively. The T7 phage library and total IgG were incubated together, and IgG-phage complexes were captured with a mixture of protein A- and G-coated magnetic beads. The beads were washed, and the captured phages were lysed, amplified by PCR, and sequenced with the NextSeq 500 platform (Illumina). Peptide-wise enrichment scores and species-specific enrichment scores were calculated as previously described ⁴⁵. Principal component analysis (PCA) was performed on species-specific enrichment scores with the FactoMineR package in R. For the heatmap analysis of peptide enrichment scores, peptides displaying no significant enrichment in any of the individuals (score < 2.3) were discarded.

Primary cells

Peripheral blood mononuclear cells (PBMCs) were isolated by the Ficoll-Hypaque density gradient centrifugation (GE Healthcare) of venous blood samples. Cells were frozen and stored at −150°C until use. Cryopreserved PBMCs were rapidly thawed in a water bath at 37 °C and washed with RPMI 1640 medium with GlutaMAX (Gibco), supplemented with 10% FBS.

Immunophenotyping of B cells by mass cytometry

Mass cytometry was performed on cryopreserved PBMCs from the PD-1-deficient patient and his healthy brother and parents in a previous study ¹⁸. B-cell subsets were analyzed in FlowJo.

Immunophenotyping of B cells by flow cytometry

Panel 1:

Freshly thawed PBMCs were stained with anti-CD5-PE, CD10-BUV737, CD19-BV711, CD20-Pacific Blue, CD21-BV421, CD23-FITC, CD27-BB515, CD38-APC, IgA-PE/Cy5, and IgG-BV605 mAbs. B cells were gated as CD20⁺ cells in FlowJo software (FlowJo) and imported into R for further analysis.

Panel 2:

Freshly thawed PBMCs were analyzed via spectral flow cytometry in a previous study ¹⁸. Briefly, cells were stained with the Zombie NIR Fixable Viability dye (BioLegend; 423106; 1:2,000) and then with anti-CD3-V450 (BD Biosciences, Clone: UCHT1, 560365; 1:450), anti-CD4-BUV563 (BD Biosciences, Clone: SK3, 612913; 1:450), anti-CD8-BUV737 (BD Biosciences, Clone: SK1, 612754; 1:150), anti-CD14-BUV395 (BD Biosciences, Clone: MφP9, 563561; 1:100), anti-CD16-PE/Dazzle 594 (BioLegend, Clone: 3G8, 302053; 1:150), anti-CD20-BV785 (BioLegend, Clone: 2H7, 302356; 1:150), anti-CD56-BV605 (BioLegend, Clone: 5.1H11, 362537; 1:50), anti-γδTCR-Alexa Fluor 647 (BioLegend, Clone: B1, 331214; 1:50), anti-Vδ1-FITC (Miltenyi Biotec, Clone: REA173, 130–118-498; 1:450), anti-Vδ2-APC/Fire 750 (BioLegend, Clone: B6, 331419; 1:3,500), anti-Vα7.2-Alexa Fluor 700 (BioLegend, Clone: 3C10, 351727; 1:50), MR1-BV421 (provided by the NIH Tetramer Core Facility; 1:200), anti-Vα24-Jα18-BV480 (BD Biosciences, Clone: 6B11, 746788; 1:50) and anti-Vβ11-APC (Miltenyi Biotec, Clone: REA559, 130–125-508; 1:150) antibodies. Cells were acquired with an Aurora Flow Cytometer (Cytek). B cells were gated as CD3⁻CD20⁺ cells in FlowJo.

Panel 3:

Freshly thawed PBMCs were analyzed with a general leukocyte immunophenotyping panel as described elsewhere ⁴⁶. Cells were acquired with an Aurora Flow Cytometer (Cytek). B cells were gated as CD3⁻CD19⁺ cells in FlowJo.

Panel 4:

Freshly thawed PBMCs were stained with the Zombie NIR Fixable Viability dye (BioLegend; 1:1000) for 15 minutes at 4 °C in the dark, washed once, and surface-stained with the following reagents for 1 hour at 4 °C in the dark: 0.1% sodium azide, BD Horizon Brilliant Stain Buffer Plus (BD Biosciences, 566385; 1:5), anti-CD16 (eBioscience, Clone: B73.1, 16–0167-82; 1:50), anti-CD32 (eBioscience, Clone: 6C4, 16–0329-81; 1:50), anti-CD11b-BUV395 (BD Biosciences, Clone: ICRF44, 563840; 1:50), anti-CD21-BUV496 (BD Biosciences, Clone: B-ly4, 750614; 1:100), anti-CD24-BUV563 (BD Biosciences, Clone: ML5, 741364; 1:50), anti-CD138-BUV615 (BD Biosciences, Clone: MI15, 751148; 1:100), anti-CD38-BUV661 (BD Biosciences, Clone: HIT2, 612969; 1:50), anti-CD10-BUV737 (BD Biosciences, Clone: HI10a, 741825; 1:100), anti-CD23-BUV805 (BD Biosciences, Clone: EBVCS-5, 749448; 1:250), anti-CXCR4-BV421 (BioLegend, Clone: 12G5, 306518; 1:250), anti-CD3-V450 (BD Biosciences, Clone: UCHT1, 560366; 1:100), anti-CD56-V450 (BD Biosciences, Clone: B159, 560361; 1:100), anti-CD5-BV480 (BD Biosciences, Clone: UCHT2, 566122; 1:50), anti-IgA-VioGreen (Miltenyi Biotec, Clone: IS11–8E10, 130–113-481; 1:100), anti-IgM-BV570 (BioLegend, Clone: MHM-88, 314517; 1:100), anti-PD-L1-BV605 (BD Biosciences, Clone: MIH1, 740426; 1:100), anti-CD19-BV650 (BD Biosciences, Clone: HIB19, 740568; 1:100), anti-PD-L2-BV711 (BD Biosciences, Clone: MIH18, 564258; 1:100), anti-B220-BV750 (BioLegend, Clone: RA3–6B2, 103261; 1:100), anti-IgD-BV785 (BioLegend, Clone: IA6–2, 348242; 1:250), anti-CD40-BB515 (BD Biosciences, Clone: 5C3, 565258; 1:250), anti-CD80-FITC (BioLegend, Clone: 2D10, 305206; 1:50), anti-CD20-Alexa Fluor 532 (eBioscience, Clone: 2H7, 58–0209-42; 1:50), anti-PD-1-BB700 (BD Biosciences, Clone: EH12.1, 566461; 1:100), anti-IgG-PE/Cy5 (BD Biosciences, Clone: G18–145, 551497; 1:20), anti-CD27-APC (BD Biosciences, Clone: M-T271, 561400; 1:20), anti-CD86-Alexa Fluor 647 (BioLegend, Clone: IT2.2, 305416; 1:250), anti-CD11c-Alexa Fluor 700 (BioLegend, Clone: Bu15, 337220; 1:1000), anti-BCMA-APC/Fire 750 (BioLegend, Clone: 19F2, 357516; 1:100), and anti-HLA-DR-APC/Fire 810 (BioLegend, Clone: L243, 307674; 1:100) antibodies. Cells were then fixed and permeabilized with the Foxp3/Transcription Factor Staining Buffer Set (Cytek, SKU TNB-0607-KIT) and stained intracellularly by overnight incubation at 4 °C with FcR blocking reagent (Miltenyi Biotec; 1:50), anti-IRF4-PE (BioLegend, Clone: IRF4.3E4, 646404; 1:100), anti-Blimp1-PE/CF594 (BD Biosciences, Clone: 6D3, 565274; 1:100), anti-T-bet-PE/Cy7 (BioLegend, Clone: 4B10, 644824; 1:500), and anti-RORγ-PerCP/eFluor 710 (eBioscience, Clone: AFKJS-9, 46–6988-82; 1:500) antibodies in permeabilization buffer. Cells were then washed three times with FACS buffer and acquired with an Aurora Flow Cytometer (Cytek). B cells were gated as CD3⁻CD56⁻CD19⁺ cells in FlowJo, with the data imported into R for further analysis. UMAP was generated from the expression levels of CD11c, CD21, CD23, IgD, IgM, and T-bet in 1,000 randomly downsampled non-plasmablast B cells (CD3⁻CD19⁺IRF4⁻Blimp1⁻) using the uwot package in R.

Fluorescence-activated cell sorting (FACS)

Freshly thawed PBMCs were stained with the following reagents in FACS buffer (2% FBS and 2 mM EDTA in PBS, filter-sterilized) for 1 hour at 4 °C in the dark: FcR blocker (Miltenyi Biotec; 1:50), anti-αβTCR-PE/Cy7 (BioLegend, Clone: IP26; 1:100), anti-CD3-FITC (Tonbo Biosciences, Clone: UCHT1; 1:100), anti-CD4-PE (BioLegend, Clone: OKT4; 1:50), anti-CD8-BV711 (BioLegend, Clone: RPA-T8; 1:50), anti-CD10-BUV737 (BD Biosciences, Clone: HI10a; 1:100), anti-CD19-BV650 (BD Biosciences, Clone: HIB19; 1:100), anti-CD20-Alexa Fluor 700 (BD Biosciences, Clone: 2H7; 1:100), anti-CD27-APC (BD Biosciences, Clone: M-T271; 1:20), anti-CD38-BV421 (BioLegend, Clone: HIT2; 1:100), anti-CD45RA-BV510 (BD Biosciences, Clone: HI100; 1:100), and anti-CCR7-BUV395 (BD Biosciences, Clone: 2-L1-A; 1:20) antibodies. Cells were then mixed with additional FACS buffer to make the final volume up to ~500 μL. We added 7-amino-actinomycin D (7-AAD; Tonbo Biosciences, 13–6993; final 1:200), filtered the cells with a Cell Strainer (Corning, 352235), and sorted the following cell subsets: naïve B cells (CD3⁻CD19⁺CD20⁺CD38⁻CD10⁻CD27⁻), memory B cells (CD3⁻CD19⁺CD20⁺CD38⁻CD27⁺), CD4⁺ αβ naïve T cells (CD19⁻CD3⁺αβTCR⁺CD4⁺CD8⁻CD45RA⁺CCR7⁺), and CD4⁺ αβ memory T cells (CD19⁻CD3⁺αβTCR⁺CD4⁺CD8⁻CD45RA⁻CCR7⁺). Cells were sorted with a FACS Aria instrument (BD Biosciences) into either 300 μL Buffer RLT Plus from the RNeasy Plus Micro Kit (Qiagen, 74034) supplemented with β-mercaptoethanol or 1 mL RPMI culture medium supplemented with gentamicin (1:500) for RNA extraction and culture, respectively. Cells were kept on ice during sorting.

Single-cell transcriptomic analysis of B cells

Single-cell RNA sequencing (scRNASeq) data for cryopreserved PBMCs from the PD-1-deficient patient (age 10 years) and his healthy brother (age 6 years) at steady state were generated in a previous study ¹⁸ (https://doi.org/10.17632/nb26v3mx3x.2). Data for the cryopreserved PBMCs from the two PD-L1-deficient patients (ages 10 and 11 years) and healthy adults and age-matched controls at steady state were generated in a previous study ¹⁹ (https://doi.org/10.17632/4fwbcswj8d.1). Single-cell trajectory analysis was performed with Monocle 3 ⁴⁷. Plasmablasts were identified by unsupervised clustering analysis with batch correction and excluded from subsequent analyses. Naïve and memory B cells were projected onto the UMAP space with Monocle 3. Differential expression between healthy controls and individuals with a deficiency of PD-1 or PD-L1 was assessed by fitting a generalized linear model with a batch variable in Monocle 3. Geneset overrepresentation analysis (GORA) was performed with the fgsea package and the Hallmark genesets retrieved from the MSigDB database (https://www.gsea-msigdb.org/gsea/msigdb/). Significantly enriched genesets downregulated in PD-1- or PD-L1-deficient cells relative to control cells are depicted. The color and number in the tiles indicate the odds ratio for geneset enrichment.

Bulk RNA sequencing (RNASeq) on PD-1-deficient naïve and memory B cells

RNA was extracted with the RNeasy Plus Micro Kit (Qiagen) from naïve or memory B cells (2 × 10⁴ cells) sorted from the PBMCs of the PD-1-deficient patient, his healthy brother, and healthy adults and age-matched controls. Full-length cDNA was generated from 1 ng total RNA with the SMART-Seq v4 Ultra Low Input RNA Kit (Clontech, 634888), and 1 ng cDNA was then used to prepare libraries with the Nextera XT DNA Library Preparation Kit (Illumina, FC-131–1024). Libraries with unique barcodes were pooled at equal molar ratios and sequenced on an Illumina NovaSeq 6000 sequencer with V1.5 reagents and NovaSeq Control Software V1.7.0 to generate 100 bp paired-end reads, according to the manufacturer’s protocol. The FASTQ files generated were first inspected by fastqc to check sequencing quality. Sequences were aligned with the GENCODE reference genome (GRCh37.p13 and GRCm38.p6 for human and mouse, respectively) with STAR aligner v2.6 ⁴⁸. Gene-level features were quantified with featureCounts v1.6.0 ⁴⁹. For principal component analysis (PCA), raw count data were normalized by variance-stabilizing transformation implemented in the DESeq2 package ⁵⁰. Differential expression (DE) analysis was performed with DESeq2. Geneset enrichment analysis (GSEA) was conducted with the fgsea package by projecting genes ranked by their fold-changes in expression with effect-size shrinkage ⁵¹ onto the Hallmark genesets retrieved from the MSigDB database (https://www.gsea-msigdb.org/gsea/msigdb/). Transcription factor (TF) activity inference analysis was performed with decoupleR ⁵² and the CollecTRI regulon database ⁵³.

Bulk RNASeq analysis of memory and antibody-secreting B cells in cancer patients on PD-1 blockade

A previously published dataset (GSE179487) ¹⁷ was analyzed with DESeq2. GSEA was conducted using the fgsea package as described above.

Bulk RNASeq on PD-L1-deficient whole-blood leukocytes

Whole-blood RNASeq data for the PD-L1-deficient individuals and their age-matched controls were generated in a previous study ¹⁹ with venous blood samples stabilized in Tempus Blood RNA tubes (Thermo Fisher Scientific) as previously described ⁵⁴. The FASTQ data were processed and analyzed as described above.

B-cell receptor (BCR) repertoire analysis

FASTQ files from the bulk RNASeq analysis of sorted naïve and memory B cells or whole-blood leukocytes were analyzed with MiXCR ⁵⁵. The physicochemical properties of productive complementarity-determining region 3 (CDR3) amino-acid sequences were calculated with the alakazam package ⁵⁶. Rarefaction analysis was performed with the iNEXT package ⁵⁷ based on the clonotype count data returned by MiXCR. Somatic hypermutation analysis was performed by parsing the MiXCR alignment results with custom R scripts.

Immunophenotyping of circulating T follicular helper (cTfh) cells

For the PD-1-deficient patient and his family members, as well as healthy age-matched and adult controls, PBMCs were stained with anti-CD3-BV570, CD20-BUV805, iNKT-BV605, γδTCR-BV711, CD4-APC, CD8-BUV496, CD25-BYG584, CD127-BB700, CD45RA-BUV395, CCR7-PE/Cy7, CXCR5-BUV615, CXCR3-BV421, CCR6-PE/CF594, and PD-1-BUV661 mAbs, as previously described ²⁷. For the two PD-L1-deficient patients and age-matched controls, as well as children with early-onset type 1 diabetes, freshly drawn venous blood leukocytes were stained with anti-CD3-APC/Cy7, CD4-BUV395, CD25-FITC, CD127-APC, CD45RA-BV786, CCR7-Pacific Blue, CXCR5-PE/Cy7, CXCR3-AmCyan, and CCR6-BUV737 mAbs. Frequencies of Tfh cells (CXCR5⁺) within non-Treg memory [central memory (Tcm, CCR7⁺CD45RA⁻), effector memory (Tem, CCR7⁻CD45RA⁻), and Temra (CD45RA⁺CCR7⁻)] CD4⁺ T cells were determined. Th1-like (CXCR3⁺CCR6⁻), Th2-like (CXCR3⁻CCR6⁻), Th17-like (CXCR3⁻CCR6⁺), and Th1*-like (CXCR3⁺CCR6⁺) phenotypes among Tfh cells were defined by the differential expression of CXCR3 and CCR6, as described elsewhere ²⁸.

Bulk RNASeq analysis of PD-1- and PD-L1-deficient leukocytes stimulated in vitro

PBMCs from the PD-1-deficient patient and his brother, the two PD-L1-deficient patients and one age-matched control, and several healthy adult controls were left non-stimulated or were stimulated with lipopolysaccharides (LPS), anti-CD2/3/28 mAb cocktail, or PMA/ionomycin for 24 hours, and bulk RNASeq was performed as described elsewhere ¹⁹. A subset of this dataset has been described elsewhere ¹⁹. The number of transcripts per million (TPM) was computed with edgeR ⁵⁸.

ProQuantum IL-21 immunoassay

Naïve or memory CD4⁺ αβ T cells were sorted, dispensed into a U-bottom plate at a density of 5 × 10⁴ cells/well in 50 μL culture medium, and allowed to rest overnight. The cells were then either left unstimulated (addition of 50 μL medium) or were stimulated with 50 μL medium supplemented with ImmunoCult^™ Human CD3/CD28/CD2 T-Cell Activator (STEMCELL, 10970, 1:100) or Cell Stimulation Cocktail (eBioscience, 00–4970-93, 1:1000) for 6 hours. We investigated whether prolonged in vitro culture rescued the defect of IL-21 production by also stimulating the sorted memory CD4⁺ αβ T cells with ImmunoCult^™ Human CD3/CD28/CD2 T-Cell Activator and culturing these cells for two weeks in ImmunoCult^™-XF T Cell Expansion Medium (STEMCELL, 10981) supplemented with recombinant human IL-2 (Roche, 10 ng/mL). Propagated T cells were counted, dispensed into a U-bottom plate at a density of 2 × 10⁵ cells/well in 100 μL culture medium, and restimulated with the same reagents for 4 hours. Supernatants were harvested and analyzed with the Human IL-21 ProQuantum Immunoassay Kit (Invitrogen, A35593).

Analysis of IL21 mRNA by RT-qPCR

Total RNA from stimulated PBMCs or sorted naïve or memory CD4⁺ αβ T cells was reverse-transcribed with SuperScript IV Reverse Transcriptase (Invitrogen, 18090050). Quantitative PCR was performed as described previously ¹⁸.

Analysis of PD-1 expression on pre-, pro-, and immature B cells in the bone marrow

Bone marrow (BM) was collected from patients who had consented to BM collection through a scheduled sternotomy during cardiac surgery. Red blood cells were removed from the collected BM in red blood cell lysis buffer, and cells were cryopreserved in 10% DMSO (Sigma-Aldrich) diluted in heat-inactivated FBS until use. After thawing, samples were stained with anti-CD10-PerCP/Cy5.5 (HI10a), CD19-APC-Alexa Fluor 700 (J3–119), CD21-FITC (B-ly4), CD27-PE (M-T271), CD34-PE/Cy7 (581), and IgM-PE/CF594 (G20–127) mAbs, together with anti-PD-1-Alexa Fluor 647 (MIH4) or the corresponding isotype control (MOPC-21) mAb, for 15 minutes at room temperature and then analyzed on a Navios flow cytometer (Beckman Coulter). Cell viability was assessed by staining with DAPI (Sigma-Aldrich). Data analysis was performed with Kaluza software (Beckman Coulter). PD-1 expression levels in pro-, pre-, and immature (CD19⁺CD27⁻CD10⁺IgM⁻CD34⁺, CD19⁺CD27⁻CD10⁺IgM⁻CD34⁻, and CD19⁺CD27⁻CD10⁺IgM⁺CD21⁻, respectively) B-cell compartments were quantified.

Analysis of PD-1 expression on B cells stimulated in vitro

B cells were sorted from PBMCs from healthy donors by negative selection with the EasyStep Human B-cell Enrichment Kit and CD43 depletion (BioLegend). B cells were plated at a density of 200,000 cells/well in a U-bottom 96-well plate, in RPMI + 10% FBS, and stimulated with 2.5 μg/mL polyclonal F(ab’)2 anti-human IgM (Jackson Immunoresearch) with 0.5 μg/mL CD40 ligand (CD40L; InvivoGen), 1 μg/mL class B CpG ODN 2006 (TLR9 agonist, InvivoGen), or 1 μg/mL resiquimod (R848; TLR7/8 agonists, InvivoGen) for 48 hours. Cells were stained with anti-PD-1-Alexa Fluor 647 mAb (Clone: MIH4) or the corresponding isotype control (MOPC-21) for 15 minutes at room temperature. Cells were co-stained with anti-CD19-PerCP/Cy5.5 (Clone: HIB19) and anti-CD27-PE (Clone: M-T271) mAbs, together with anti-CD86-FITC mAb (Clone: 2331 (FUN-1)) as a marker for B-cell activation. Cells were then analyzed on a Navios flow cytometer (Beckman Coulter). Cell viability was assessed by staining with DAPI (Sigma-Aldrich). Data analysis was performed with Kaluza software (Beckman Coulter). PD-1 expression levels in gated CD19⁺, CD19⁺CD27⁻ (naïve), and CD19⁺CD27⁺ (memory) B-cell compartments were quantified.

Analysis of PD-L1 expression on B cells among stimulated PBMCs

PD-L1 expression on stimulated PBMCs has been described elsewhere ¹⁹. Briefly, PBMCs from the two PD-L1-deficient patients and their age-matched control, and one adult control, were stimulated with IFN-α2, LPS, or bead-conjugated anti-CD3/28 mAbs for 24 hours. PD-L1 expression was quantified by flow cytometry.

We investigated the impact of in vitro IFN-γ neutralization on the induction of PD-L1 expression in B cells among PBMCs, using anti-IFN-γ mAb (eBioscience, 16–7318-81, clone NIB42, 10 μg/mL) or mouse IgG1 isotype control (MOPC-21). Cells were stained with the following reagents in FACS buffer for 1 hour at 4 °C in the dark: anti-CD21-BUV496 (BD Biosciences, Clone: B-ly4; 1:100), anti-CD24-BUV563 (BD Biosciences, Clone: ML5; 1:50), anti-CD138-BUV661 (BD Biosciences, Clone: MI15; 1:100), anti-CD38-BUV661 (BD Biosciences, Clone: HIT2; 1:50), anti-CD10-BUV737 (BD Biosciences, Clone: HI10a; 1:100), anti-CD23-BUV805 (BD Biosciences, Clone: EBVCS-5; 1:250), anti-CD8-BV421 (BioLegend, Clone: RPA-T8, 1:100), anti-CD56-V450 (BD Biosciences, Clone: B159, 1:50), anti-CD123-BV480 (BD Biosciences, Clone: 9F5, 1:100), anti-IgA-VioGreen (Miltenyi Biotec, Clone: IS11–8E10, 1:100), anti-IgM-BV570 (BioLegend, Clone: MHM-88, 1:100), anti-CD19-BV650 (BD Biosciences, Clone: HIB19; 1:100), anti-PD-L1-BV711 (BioLegend, Clone: 29E.2A3, 1:100), anti-IgD-BV785 (BioLegend, Clone: IA6–2, 1:250), anti-CD3-FITC (Tonbo Biosciences, Clone: UCHT1, 1:100), anti-PD-1-BB700 (BD Biosciences, Clone: EH12.1, 1:100), anti-CD4-PE (BD Biosciences, Clone: OKT4, 1:100), anti-CD16-PE/Dazzle 594 (BioLegend, Clone: 3G8, 1:100), anti-IgG-PE/Cy5 (BD Biosciences, Clone: G18–145, 1:20), anti-CD20-PE/Cy7 (BioLegend, Clone: 2H7, 1:100), anti-CD27-APC (BD Biosciences, Clone: M-T271; 1:20), anti-CD14-Spark NIR 685 (BioLegend, Clone: 63D3; 1:100), anti-CD11c-Alexa Fluor 700 (BioLegend, Clone: Bu15; 1:1000), and anti-HLA-DR-APC/Fire 810 (BioLegend, Clone: L243; 1:100) antibodies. After washing with FACS buffer, cells were resuspended with FACS buffer containing 7-AAD (Tonbo Biosciences, 1:200) and acquired with an Aurora Flow Cytometer (Cytek). A Fluorescence Minus One (FMO) control for the anti-PD-L1-BV711 antibody was also prepared to determine the background signal.

Analysis of PD-1, PD-L1, and PD-L2 expression on sorted naïve B cells stimulated in vitro

Naïve B cells from healthy donors were sorted as CD20⁺CD10⁻CD27⁻IgG⁻ cells and cultured (50,000/200 μL/well) in B-cell medium alone or in the presence of recombinant HA-tagged CD40L (R & D Systems, 200 ng/mL) plus anti-HA mAb (R & D Systems, 50 ng/mL), CpG 2006 (InvivoGen, 1 μg/mL), F(ab’)₂ goat anti-human IgM/G/A (aBCR; Jackson ImmunoResearch Labs, 2.5 μg/mL), recombinant human IL-21 (PeproTech, 50 ng/mL) or combinations of these stimuli. After 48 hours, the cells were harvested and stained with anti-PD-1-BV605 (BD Biosciences, Clone EH12.1), anti-PD-L1-PE (BioLegend, Clone 29E.2A3), and anti-PD-L2-APC (BioLegend, Clone 24F.10C12) antibodies. Data were acquired on a BD LSRII Fortessa (BD Biosciences) and analyzed with FlowJo. Geometric mean fluorescence intensity (gMFI) was computed and then normalized by the mean of gMFI in the non-stimulated samples.

Validation of the PD-1-neutralizing capacity of anti-PD-1 biosimilars

HuT78 and Raji cell coculture assay was performed, as previously described ¹⁸, to validate the neutralizing activity of commercially available biosimilars for nivolumab and pembrolizumab. Briefly, HuT78 cells (transduced with EV or WT PD-1; 1 × 10⁵ cells per well) and Raji cells (transduced with EV, PD-L1, or PD-L2; 1 × 10⁵ cells per well) were cocultured for 24 hours with or without a bispecific antibody directed against CD19 and CD3 (equivalent to blinatumomab) (BPS Bioscience, 100441–2, 10 ng/mL), and anti-PD-1-hIgG4 antibody (S228P) (InvivoGen, Cat: hpd1ni-mab114 or hpd1pe-mab14, 5 μg/mL) or isotype control (InvivoGen, Cat: bgal-mab114). Monensin and brefeldin A (Cytek, 1:1000 each) were added for the last six hours of incubation. Cells were stained with Zombie NIR Fixable Viability dye (BioLegend, 1:1000 in PBS) for 15 minutes at 4 °C in the dark, washed with FACS buffer, and fixed and permeabilized with the Foxp3/Transcription Factor Staining Buffer Set (Cytek). Cells were then stained by incubation overnight at 4 °C in the dark with the following reagents in permeabilization buffer: FcR blocking reagent (Miltenyi Biotec, 1:50), anti-CD3-APC (Cytek, Clone: UCHT1, 1:100), anti-IFN-γ-PE-Dazzle 594 (BioLegend, Clone: 4S.B3, 1:500), and anti-TNF-BV711 (BioLegend, Clone: MAb11, 1:500) mAbs. The cells were washed with FACS buffer and acquired with an Attune NxT Flow Cytometer with a CytKick MAX Autosampler (Invitrogen). Data were analyzed with FlowJo and R. The percentage of IFN-γ⁺ cells was used as a readout.

Analysis of proliferation and Ig secretion by naïve B cells stimulated in vitro

Naïve (CD20⁺CD10⁻CD27⁻IgG⁻ or CD3⁻CD19⁺CD20⁺CD38⁻CD10⁻CD27⁻) B cells were sorted from the PBMCs of the PD-1-deficient patient, his brother, or healthy age-matched and adult controls. Previously reported data for a patient with c-Rel deficiency were included as a disease control ³⁰. Data for additional disease controls [STAT3 GOF patients (N=4), patients with NF-κB2 haploinsufficiency (N=3), and a patient with RelB deficiency] were also included. The purity of the recovered populations was typically >98%. For proliferation assays, cells were stained with CFSE dye (Cytek, SKU 13–0850-U500, 1:10000) at room temperature in the dark for 10 minutes. B cells were cultured in 96-well U-bottom plates (5 × 10³ cells in 200 μL culture medium) for 7 days with HA-tagged CD40L (R&D Systems, 200 ng/mL) plus anti-HA mAb (R&D Systems, 50 ng/mL) wither with or without IL-21 (PeproTech, 50 ng/mL) or CpG 2006 (InvivoGen, 1 μg/mL). For in vitro PD-1 blockade assays, anti-PD-1-hIgG4 antibody (S228P) (InvivoGen, Cat: hpd1ni-mab114 or hpd1pe-mab14, 500 ng/mL) or isotype control (InvivoGen, Cat: bgal-mab114) was added at the start of culture. Mouse anti-human PD-1 mAb (GenScript, Clone PD1.D3, 5 μg/mL) or mouse IgG1 isotype control was also tested. Culture supernatants were harvested after 7 days, and the amounts of IgM, IgA, and IgG secreted into the supernatant were determined by Ig heavy chain-specific ELISA or a LEGENDplex Human Immunoglobulin Isotyping assay (BioLegend, Cat: 740638).

Analysis of proliferation and Ig secretion by memory B cells stimulated in vitro

Memory (CD3⁻CD19⁺CD20⁺CD38⁻CD27⁺) B cells were isolated from the PBMCs of the PD-1-deficient patient, his brother, or healthy age-matched and adult controls by sorting. B cells were cultured in 96-well U-bottom plates (100 cells in 50 μL medium) for 14 days in ImmunoCult^™ Human B-Cell Expansion medium (STEMCELL, Cat: 100–0645) with the supplements added immediately before use. After 7 days, 50 μL of fresh medium with the supplements was added to the culture. Culture supernatants were harvested after 14 days, and the amounts of IgM, IgG, and IgA secreted into the supernatant were determined in a LEGENDplex Human Immunoglobulin Isotyping assay (BioLegend). Cells within the wells were resuspended in FACS buffer and 7-AAD (Tonbo Biosciences, 1:200), and acquired with an Attune NxT Flow Cytometer with a CytKick MAX Autosampler (Invitrogen), with CountBright^™ Absolute Counting Beads (Invitrogen, C36950) for accurate counting.

Bulk RNASeq analysis of naïve B cells stimulated in vitro

Naïve (CD3⁻CD19⁺CD20⁺CD38⁻CD10⁻CD27⁻) B cells were sorted, dispensed into a U-bottom plate at a density of 1 × 10⁴ cells/well in 50 μL culture medium, and allowed to rest overnight. The cells were then stimulated with MEGACD40L (Enzo, ALX-522–110-C010, 200 ng/mL) plus IL-21 (PeproTech, 200–21, 50 ng/mL) or AffiniPure F(ab’)₂ Fragment Goat Anti-Human IgA + IgG + IgM (H+L) (Jackson ImmunoResearch, 109–006-064, 0.83 μg/mL) plus CpG ODN 2006 (InvivoGen, tlrl-2006, 1 μg/mL) for 4, 24, or 48 hours. RNA was extracted and processed for RNASeq as described above. Weighted gene co-expression network analysis (WGCNA) was performed in R ³². A soft threshold of 10 was chosen to obtain a signed R² > 0.9. Gene modules that were differentially induced or suppressed were identified by applying the limma differential analysis framework to the eigengene matrix module. A hypergeometric geneset overrepresentation test was performed with hypeR ⁵⁹ by projecting the genes assigned to each module onto MSigDB genesets. For in vitro PD-1 blockade assay, anti-PD-1-hIgG4 antibody (S228P) (InvivoGen, Cat: hpd1ni-mab114 or hpd1pe-mab14, 5 μg/mL) or isotype control (InvivoGen, Cat: bgal-mab114) was added at the start of culture. DE analysis was performed with limma-voom ⁶⁰. GSEA was performed with the fgsea package and the Hallmark genesets retrieved from the MSigDB database (https://www.gsea-msigdb.org/gsea/msigdb/).

Analysis of c-Myc expression in B cells

Freshly thawed PBMCs or sorted naïve or memory B cells stimulated for 7 days were stained with the Ghost Dye Red 780 (Cytek, 1:10000 in PBS) for 15 minutes at 4 °C in the dark, washed once, and surface-stained with the following reagents for 1 hour at 4 °C in the dark: 0.1% sodium azide, BD Horizon Brilliant Stain Buffer Plus (BD Biosciences, 1:5), anti-CD16 (eBioscience, Clone: B73.1, 1:50), anti-CD32 (eBioscience, Clone: 6C4, 1:50), anti-CD21-BUV496 (BD Biosciences, Clone: B-ly4, 1:100), anti-CD24-BUV563 (BD Biosciences, Clone: ML5, 1:50), anti-CD138-BUV615 (BD Biosciences, Clone: MI15, 1:100), anti-CD38-BUV661 (BD Biosciences, Clone: HIT2, 1:50), anti-CD10-BUV737 (BD Biosciences, Clone: HI10a, 1:100), anti-CD23-BUV805 (BD Biosciences, Clone: EBVCS-5, 1:250), anti-CXCR4-BV421 (BioLegend, Clone: 12G5, 1:250), anti-CD3-V450 (BD Biosciences, Clone: UCHT1, 1:100), anti-CD5-BV480 (BD Biosciences, Clone: UCHT2, 1:50), anti-IgA-VioGreen (Miltenyi Biotec, Clone: IS11–8E10, 1:100), anti-IgM-BV570 (BioLegend, Clone: MHM-88, 1:100), anti-CD19-BV650 (BD Biosciences, Clone: HIB19, 1:100), anti-IgD-BV785 (BioLegend, Clone: IA6–2, 1:250), anti-CD40-BB515 (BD Biosciences, Clone: 5C3, 1:250), anti-IgG-PE/Cy5 (BD Biosciences, Clone: G18–145, 1:20), anti-CD27-APC (BD Biosciences, Clone: M-T271, 1:20), and anti-CD20-Alexa Fluor 700 (BD Biosciences, Clone: 2H7, 1:100) antibodies. Cells were then fixed and permeabilized with the Foxp3/Transcription Factor Staining Buffer Set (Cytek) and stained intracellularly by overnight incubation at 4 °C with FcR blocking reagent (Miltenyi Biotec; 1:50), anti-IRF4-PE (BioLegend, Clone: IRF4.3E4, 1:500), anti-Blimp1-PE/CF594 (BD Biosciences, Clone: 6D3, 1:500), anti-T-bet-PE/Cy7 (BioLegend, Clone: 4B10, 1:500), anti-RORγ-PerCP/eFluor 710 (eBioscience, Clone: AFKJS-9, 1:500), and anti-c-Myc-Alexa Fluor 647 (Santa Cruz Biotechnology, sc-42 AF647, Clone: C-33, 1:100) antibodies in permeabilization buffer. Cells were then washed three times with FACS buffer and acquired with an Aurora Flow Cytometer (Cytek).

Pharmacological inhibition of SHP2 and Myc

SHP2 inhibitors (SHP099, RMC-4550) and Myc inhibitors (10058-F4, EN4, and KJ Pyr 9) were purchased from MedChemExpress and reconstituted in DMSO to make 10 mM stock solutions. Inhibitors were tested at 1 or 50 μM. The corresponding volume of DMSO was used as a vehicle control. Six technical replicates were prepared for each condition in the memory B-cell expansion assay. Secreted Ig levels were averaged between technical replicates.

Construction of a Pdcd1^fl/fl mb1-Cre mouse line

Pdcd1 floxed mice (Pdcd1^fl/fl; CDB1220K) were generated as previously described ⁶¹. Briefly, a genomic fragment of exons 2 and 3 of the Pdcd1 locus was flanked by loxP sites, together with a neomycin resistance gene cassette (PGK-Neo-pA) flanked by FLP recombinase target (FRT) sequences, to construct a targeting vector. The targeting vector was introduced into HK3i embryonic stem cells (ESCs) derived from C57BL/6 mice ⁶¹. The homologous recombinant ESCs were identified by screening, and ESC clones were microinjected into eight-cell stage ICR embryos. The embryos were transferred into pseudopregnant ICR female mice, and germline transmission of the targeted Pdcd1 allele was observed in the resulting chimeric mice. The chimeric mice were first crossed with C57BL/6 mice. The heterozygous offspring of this cross were then crossed with CAG-FLPe deleter mice ⁶² to establish the Pdcd1 floxed mice. The primers used for Pdcd1 genotyping were as follows: Fwd (5’-GCTGCTGGCTTTCCTAAACA-3’) and Rev (5’-TGTCAGGCACTGAAGAGATCTACAC-3’) for the WT allele (273 bp) and the floxed allele (439 bp). Cd79a (mb1) Cre/⁺ mice (C57BL/6) were provided by Michael Reth (University of Freiburg) ⁴⁴. Cd79a (mb1) Cre/⁺ and Pdcd1 floxed mice were crossed to establish the Pdcd1^fl/fl mb1-Cre mice. After stimulating peripheral blood leukocytes (PBLs) with LPS (1 μg/mL) for two days in vitro, PD-1 expression on CD19⁺B220⁺ B cells and non-B cells was quantified by flow cytometry.

Flow cytometric immunophenotyping of B cells in Pdcd1^−/− and Pdcd1^fl/fl mb1-Cre mice

Young (3–4 months old) and aged (10–12 months old) female WT or Pdcd1^−/− C57BL/6 mice ⁴ and young (3–4 months old) or aged (13–14 months old) Flox ctrl (Pdcd1^fl/fl) or Pdcd1^fl/fl mb1-Cre C57BL/6 mice were analyzed. Bilateral axillary, brachial, and inguinal LNs were pooled and homogenized. Spleens and bone marrow were harvested, homogenized, and treated with ammonium chloride potassium buffer for 2 minutes to lyse red blood cells. Absolute cell counts were calculated per LN, spleen, or bone marrow specimen. For flow cytometric analysis, 1~5 × 10⁵ cells were stained with the Zombie NIR Fixable Viability dye (BioLegend) and Fc-blocked (BD Biosciences, Cat: 553142; Clone: 2.4G2; 1:200). For surface staining, cells were stained with diluted antibodies for 12 minutes at 4 °C in the dark. The following reagents were used: anti-CD19-BUV395 (BD Biosciences, Cat: 563557, Clone: 1D3; 1:200), anti-CD45.2-BV786 (BD Biosciences, Cat: 563686, Clone: 104; 1:200), anti-CD45.2-BUV737 (BD Biosciences, Cat: 612778, Clone: 104; 1;200), anti-CD45R(B220)-PE (eBioscience, Cat: 12–0452-81, Clone: RA3–6B2; 1;200), anti-CD45R(B220)-FITC (Tonbo Biosciences, Cat: 35–0452, Clone: RA3–6B2; 1:200), anti-CD45R(B220)-BV785 (BioLegend, Cat: 103246, Clone: RA3–6B2; 1;200), anti-CD80-BV510 (BD Biosciences, Cat: 740130, Clone: 16–10A1; 1:200), anti-CD80-BUV737 (BD Biosciences, Cat: 612773, Clone: 16–10A1; 1;200), anti-CD73-PE (BioLegend, Cat: 127206, Clone: TY/11.8; 1;200), anti-CD73-PerCP/Cy5.5 (BioLegend, Cat: 127213, Clone: TY/11.8; 1:200), anti-CD3-APC/Cy7 (BioLegend, Cat: 100222, Clone: 17A2; 1;200), anti-F4/80-APC/Cy7 (BioLegend, Cat: 123118, Clone: BM8; 1;200), anti-CD43-PE (BD Biosciences, Cat: 553271, Clone: S7; 1;200), anti-IgM-PE/Cy7 (eBioscience, Cat: 25–5790-82, Clone: II/41; 1;200), anti-Ly-51(CD249)-FITC (BioLegend, Cat: 108305, Clone: 6C3; 1;200), anti-IgD-Pacific Blue (BioLegend, Cat: 405712, Clone: 11–26c.2a; 1;200), anti-CD24-BV510 (BioLegend, Cat: 101831, Clone: M1/69; 1;200), anti-PD-1-APC (BioLegend, Cat: 109112, Clone: RMP1–30; 1;200), anti-PD-1-PE/Cy7 (BioLegend, Cat: 109110, Clone: RMP1–30; 1;200), anti-PD-1-BUV395 (BD Biosciences, Cat: 568595, Clone: 29F.1A12; 1;200), anti-CD23-PE/Cy7 (BioLegend, Cat: 101613, Clone: B3B4; 1;200), anti-CD21/CD35(CR2/CR1)-FITC (BioLegend, Cat: 123407, Clone: 7E9; 1;200), anti-CD11c-BV421 (BD Biosciences, Cat: 562782, Clone: HL3; 1;200), anti-CD95(Fas)-APC (eBioscience, Cat: 17–0951-82, Clone: 15A7; 1:200), anti-CD83-PE (BioLegend, Cat: 121508, Clone: Michel-19; 1:200), anti-CD38-PE/Cy7 (BioLegend, Cat: 102718, Clone: 90; 1:200), anti-CXCR4-BV421 (BioLegend, Cat: 146511, Clone: L276F12; 1:200), anti-CD86-BV510 (BD Biosciences, Cat: 563077, Clone: GL1; 1:200), mouse IgG1 isotype control-APC (Tonbo Biosciences, Cat: 20–4714, Clone: MOPC-21), rat IgG1 k isotype control-PE (MBL, Cat: M080–5, Clone: 1H5), rat IgG2a k isotype control-PE-Cy7 (BioLegend, Cat: 400522, Clone: RTK2758) and rat IgG2b, κ Isotype Ctrl-APC (BioLegend, Cat: 400611, Clone: RTK4530; 1;200) antibodies. Cells were washed with PBS plus 2% FBS. For intracellular staining, cells were fixed and permeabilized with the Foxp3/Transcription Factor Staining Buffer Set (eBioscience). Cells were then stained by incubation with diluted antibodies in permeabilization buffer in the dark for 30 minutes at 4 °C. The following reagents were used: anti-T-bet-APC (BioLegend, Cat: 644813, Clone: 4B10; 1;200), anti-T-bet-APC (Miltenyi Biotec, Cat: 130–119-783, Clone: REA102; 1;200) and anti-Ki67-BUV395 (BD Biosciences, Cat: 564071, Clone: B56; 1;200), anti-c-Myc-FITC (Santa Cruz Biotechnology, Cat: sc-42 FITC, Clone: C-33; 1:200), human IgG1 isotype control-APC (Miltenyi Biotec, Cat: 130–120-709, Clone: REA293) and mouse IgG1 isotype control-FITC (Tonbo Biosciences, Cat: 35–4714, Clone: MOPC-21) antibodies. Cells were washed with PBS plus 2% FBS and acquired with an LSRFortessa X-20 (BD Biosciences). The data were analyzed in FlowJo. Memory B cells were gated as CD80⁺CD73⁺ B cells, as previously described ²³. Germinal center (GC; CD38^loFas⁺) B cells – both dark (CXCR4^hiCD86^loCD83^lo) and light zone (CXCR4^loCD86^hi CD83^hi) – were gated as previously described ²⁴.

Analysis of memory B cells after repeated oral exposure to microbial antigens in Pdcd1^−/− mice

Young (7 weeks old) male WT or Pdcd1^−/− C57BL/6 mice were treated orally with a cocktail of heat-killed E.coli 0111:B4 (HKEB; InvivoGen, cat: tlrl-hkeb2, 1×10⁶ cells per mouse), heat-killed Mycobacterium tuberculosis (HKMT; InvivoGen, cat: tlrl-hkmt-1, 2μg per mouse) and beta-1,3-glucan from Alcaligenes faecalis (Curdlan; InvivoGen, cat: tlrl-curd, 0.1mg per mouse) in PBS twice weekly. After repeated oral exposure to microbial antigens for 6 weeks, bone marrow, LN and splenic cells from these mice were analyzed; 1 × 10⁶ cells were stained with the Zombie NIR Fixable Viability dye (BioLegend) and Fc-blocked (BD Biosciences, Cat: 553142; Clone: 2.4G2; 1:200). Cells were stained by incubation with diluted anti-IgA-APC (eBioscience, Cat: 17–4204-82, Clone: mA-6E1; 1:200) antibody for 12 minutes at 4 °C in the dark, washed once, and stained with the following reagents for 12 minutes at 4 °C in the dark, anti-IgM-PE/Cy7 (eBioscience, Cat: 25–5790-82, Clone: II/41; 1;200), anti-CD73-PerCP/Cy5.5 (BioLegend, Cat: 127213, Clone: TY/11.8; 1:200), anti-CD45R(B220)-BV785 (BioLegend, Cat: 103246, Clone: RA3–6B2; 1;200), anti-CD19-BUV395 (BD Biosciences, Cat: 563557, Clone: 1D3; 1:200), anti-CD80-BUV737 (BD Biosciences, Cat: 612773, Clone: 16–10A1; 1;200) antibodies. Cells were washed with PBS plus 2% FBS and acquired with an LSRFortessa X-20 (BD Biosciences).

Analysis of recall Ab responses after sequential immunization in mice deficient for PD-1 signaling

Mice were immunized intraperitoneally with 50 μg alum-precipitated NP-CGG to initiate a primary Ab response, as previously described ⁷. Secondary responses were initiated by immunizing mice intraperitoneally with NP-CGG in PBS 12–15 weeks after the primary immunization. Antigen-specific (NP-binding) IgG1⁺ antibody-secreting plasma cells were quantified with an ELISPOT assay, as previously described ⁷. A subset of data points for day 0 (prior to the second immunization) has been published ⁷. For the time-course analysis of Pdcd1^−/− mice following the second immunization, statistical significance was assessed only for time points for which data were available for at least 7 mice.

Bulk RNASeq analysis of naïve and memory B cells in Pdcd1^−/− mice

LN or splenic cells from aged (11 months old) WT or Pdcd1^−/− C57BL/6 mice were analyzed. Cells were stained with 7-amino-actinomycin D (7-AAD; eBioscience, Cat: 00–6993-50, 1;80) and Fc-blocked (BD Biosciences, Cat: 553142; Clone: 2.4G2; 1:200). Cells were stained with the following reagents for 12 minutes at 4 °C in the dark: anti-CD3-APC (BioLegend, Cat: 100236, Clone: 17A2; 1;200), anti-CD73-PE (BioLegend, Cat: 127206, Clone: TY/11.8; 1;200), anti-CD19-FITC (Tonbo Biosciences, Cat: 35–0193, Clone: 1D3; 1;200), anti-CD80-BV510 (BD Biosciences, Cat: 740130, Clone: 16–10A1; 1;100) and anti-CD45R(B220)-BV785 (BioLegend, Cat: 103246, Clone: RA3–6B2; 1;200) antibodies. Cells were washed with PBS plus 2% FBS and sorted with a FACS Melody Cell Sorter (BD Biosciences). Memory B cells were gated as the CD80⁺CD73⁺ B-cell compartment, whereas the CD73⁻ B-cell compartment was collected as naïve B cells. For LN, cells from three mice of identical genotypes were pooled due to the small number of cells collected. Bulk RNASeq analysis was performed as described above. GSEA was performed by projecting the log₂ fold-change ranks of genes in Pdcd1^−/− mice relative to WT mice onto the Hallmark genesets. Genesets displaying enrichment or depletion, with false-discovery rate (FDR)-adjusted P values < 10⁻⁵ are depicted.

Analysis of IL-21 production by memory CD4⁺ T cells in Pdcd1^−/− mice

LN and splenic cells from aged (10 months old) WT or Pdcd1^−/− C57BL/6 mice (N=10 and 9, respectively) were analyzed. LN or splenic cells from mice of the same genotype were pooled, and CD4⁺ T cells were enriched with CD4(L3T4) MicroBeads (Miltenyi Biotec, Cat: 130–117-043). Memory CD4⁺ T cells were then isolated by FACS. In brief, cells were stained with 7-amino-actinomycin D (7-AAD; eBioscience, Cat: 00–6993-50, 1;80), Fc-blocked (BD Biosciences, Cat:553142; Clone: 2.4G2; 1:200), and stained by incubation with the following reagents for 12 minutes at 4 °C in the dark: anti-CD44-APC (BioLegend, Cat:103011, Clone: IM7; 1:200), anti-CD4-PE-Cy7 (eBioscience, Cat:25–0042-82, Clone: RM4–5; 1:200), and anti-CD45RB-BV510 (BD Biosciences, Cat: 740107, Clone: 16A; 1:200) antibodies. Cells were washed with PBS plus 2% FBS and sorted with a FACS Melody Cell Sorter (BD Biosciences). Memory CD4⁺ T cells were gated as the CD44^highCD45RB^lowCD4⁺ T-cell compartment. Isolated memory CD4⁺ T cells were stimulated with anti-CD3/CD28 Dynabeads (Gibco, Cat:11453D) for 2 days and re-stimulated with PMA (Sigma-Aldrich), ionomycin (Sigma-Aldrich), and brefeldin A (eBioscience, Cat: 00–4506-51, 1:1000) for 4 hours.

Cells with or without stimulation were stained with Zombie NIR Fixable Viability dye (BioLegend) and Fc-blocked (BD Biosciences, Cat: 553142; Clone: 2.4G2; 1:200). Cells were stained by incubation with the following reagents for 12 minutes at 4 °C in the dark: anti-CD44-FITC (BD Bioscience, Cat: 561859, Clone: IM7; 1;200), anti-CD45RB-BV510 (BD Biosciences, Cat: 740107, Clone: 16A; 1:200), anti-CD4-PE-Cy7 (eBioscience, Cat:25–0042-82, Clone: RM4–5; 1:200), anti-CD4-BV786 (BD Biosciences, Cat: 740873, Clone: H129.19; 1:200), anti-CD45.2-BUV395 (BD Biosciences, Cat: 564616, Clone: 104; 1:200), and anti-CD45.2-BUV737 (BD Biosciences, Cat: 612778, Clone: 104; 1:200) antibodies and washed with PBS plus 2% FBS. For intracellular IL-21 staining, cells were fixed and permeabilized with the Foxp3/Transcription Factor Staining Buffer Set (eBioscience). Cells were then stained by incubation with diluted anti-IL-21-PE (eBioscience, Cat: 12–7213-82, Clone: mhalx21; 1;200) antibody in permeabilization buffer for 30 minutes at 4 °C in the dark. Cells were washed with PBS plus 2% FBS and acquired with a FACSymphony^™ A5 (BD Biosciences).

QUANTIFICATION AND STATISTICAL ANALYSIS

All statistical analyses were performed in R v4 (http://www.R-project.org/) ⁶³. The statistical significance of quantitative differences between groups was assessed in two-tailed unpaired Wilcoxon’s rank-sum tests unless otherwise stated. False discovery rate (FDR) adjustment was performed via the Benjamini and Hochberg method ⁶⁴. P values below 0.05 were considered statistically significant.

Highlights.

• Impaired B cell memory in PD-1- and PD-L1-deficient humans and mice deficient in PD-1 signaling

• PD-1 deficiency, but not PD-L1 deficiency, disrupts Tfh cell phenotype and function

• Human PD-1:PD-L1 axis promotes antibody responses in a B-cell-intrinsic manner

• B cell-specific deletion of PD-1 triggers striking phenotypic alterations in mice